# Connect with Slack Source: https://docs.genloop.ai/integrate-and-extend/connect-with-slack Install the Genloop Slack app to ask data questions in any channel, get verified answers inline, and share insights with your team without leaving Slack. Connecting Genloop to Slack lets your team ask data questions directly from any Slack channel and get verified answers without leaving the conversation. ## Setting up the Slack integration Go to **Connections** and click the **Interfaces** tab. Click **Connect** on the Slack option. This opens the Slack authorization page — approve the connection and Genloop is linked to your workspace. Once connected, you can mention Genloop in any channel to ask a question. Answers come back in the thread with the same reasoning and verification as Chat. \[Screenshot: Interfaces tab showing Slack connection] If you run into any issues during setup, [talk to us](https://cal.com/ayush-genloop). ## Next steps Connect Genloop to Claude and other AI tools via MCP. Ask questions directly on the platform. # Embed via API & SDK Source: https://docs.genloop.ai/integrate-and-extend/embed-via-api-sdk Embed Genloop chat into your own applications and query your data warehouse programmatically with the REST API, SDKs, and authentication tokens. The Genloop API and SDK are currently in development. This page will be updated with full documentation once available. If you want early access or have a specific use case, [talk to us](https://cal.com/ayush-genloop). The Genloop API lets you query your data programmatically and embed verified answers into your own applications — internal tools, customer-facing products, or automated workflows. API access is available on **Pro** and above. ## What you will be able to do * Query your Genloop workspace programmatically via REST API * Embed data answers into internal or external applications * Trigger automations and retrieve Liveboard data * Use iFrames for embedded interfaces (Enterprise) ## Next steps Connect Genloop to AI tools via MCP while API docs are in progress. Get early API access or discuss your integration needs. # Genloop MCP Source: https://docs.genloop.ai/integrate-and-extend/genloop-mcp Connect Genloop to Claude, Cursor, and other AI assistants as a verified data source using the Model Context Protocol (MCP) for safe data queries. Genloop supports the Model Context Protocol (MCP), which lets you connect Genloop as a verified data source for AI tools like Claude. Once connected, any AI tool that supports MCP can query your data through Genloop — getting the same verified, context-aware answers your team gets in Chat. ## Setting up MCP **On the platform** Go to **Connections** and click the **MCP** tab. Click **Generate Token** to create a unique access token for your workspace. Copy this token — you will need it when connecting from your AI tool. You can revoke and regenerate a new token at any time from the same page. \[Screenshot: MCP tab showing token generation] **In Claude** Open Claude and find the Genloop option in the MCP integrations panel. Click it and follow the authorization flow. Once authorized, Claude can query your Genloop workspace directly. MCP setup in Claude is currently in progress. If you run into any issues during setup, [talk to us](https://cal.com/ayush-genloop) and we will help you get connected. ## Next steps Bring Genloop answers into your Slack workspace. # Roles & permissions Source: https://docs.genloop.ai/manage-workspace/roles-and-permissions Govern workspace access with Genloop's built-in Admin, Editor, and Viewer roles, or create custom roles to fine-tune permissions for data and connections. Genloop uses role-based access control to define what each member of your workspace can see and do. Roles are assigned per member and can be updated at any time from Workspace settings. ## Built-in roles Genloop ships with four built-in roles that cover most team structures: **Admin** Full access to everything in the workspace. Can manage members, edit connections, update Context Hub, review and approve in Review Center, and manage billing. **Data Manager** Can manage data connections, edit Context Hub, and review validation requests in Review Center. Cannot manage members or billing. **Analyst** Can ask questions in Chat, build and edit Liveboards, create reports, and set up Automations. Cannot manage connections, Context Hub, or members. **Viewer** Read-only access. Can view Liveboards and Chat answers but cannot ask new questions or make changes. ## Custom roles Custom roles are available on the Growth and Enterprise plans. On Free and Pro, you can only use the four built-in roles. On Growth and Enterprise, you can create custom roles with a specific combination of permissions. Custom roles let you scope access at the data source, table, Liveboard, column, and row level — so analysts, executives, and admins all work within the exact boundaries you set. Enterprise plans also support custom role auto-sync with your identity provider via SSO and SAML. ## Assigning roles Roles are assigned when you invite a member or can be updated from the members list in Workspace settings. Click the role dropdown next to any member to change it. ## Next steps Invite and manage your team. Upgrade to Growth or Enterprise for custom roles. # Workspace & members Source: https://docs.genloop.ai/manage-workspace/workspace-and-members Create and configure your Genloop workspace, invite team members by email or domain, assign roles, and manage seats from one central settings page. Your workspace is the shared environment where your team connects data, asks questions, and builds Liveboards together. The Workspace settings page is where you manage members, control access, and keep your setup organised. Go to **Workspace** in the left sidebar to access settings. ## Inviting members Click **Invite Members** to add someone to your workspace. Enter their work email and they will receive an invitation to join. Once they accept, they are part of your workspace and can access Chat, Liveboards, and connected data based on their assigned role. The number of members you can add depends on your plan. Free allows up to 2 members, Pro up to 20, Growth up to 200, and Enterprise has no limit. ## Managing members The members list shows everyone currently in your workspace with their role, email, and join date. From here you can: * Change a member's role * Remove a member from the workspace * Resend a pending invitation ## Workspace switcher If you are part of multiple workspaces, you can switch between them from the workspace switcher in the top left of the sidebar. Each workspace has its own data connections, context, and chat history. ## Next steps Control what each member can see and do in your workspace. View your credit balance and workspace limits. # How Genloop Credits Work Source: https://docs.genloop.ai/plans/credits One credit equals one data query. Credits pool across your workspace, refresh monthly, and are never consumed by idle connections or inactive members. Credits are Genloop's unit of consumption. Every time you or a teammate sends a data question to Genloop — whether from the app, Slack, or the API — it uses one credit. Credits are shared across your entire workspace, refresh at the start of each billing month, and are never consumed by anything other than active queries. ## What counts as one credit One credit is consumed when Genloop performs a data investigation or query on your behalf. This includes: * A question asked in the conversational analytics interface * A query triggered via Automations * An API call that returns a data analysis * A Slack message that triggers a query The following actions do **not** consume credits: * Viewing or refreshing a liveboard * Browsing your data source schema * Editing workspace settings * Members being idle or logged in without querying ## Credits per plan | Plan | Credits per month | | -------------- | ----------------- | | **Free** | 25 | | **Pro** | 100 | | **Growth** | 1,000 | | **Enterprise** | Custom | All credits are pooled across every member in your workspace. If your workspace has 10 members on the Pro plan, those 10 members share 100 credits per month — any member can use them. Enterprise plans include a custom credit allocation negotiated as part of your contract. Contact your Solutions Architect to discuss your expected usage volume. ## Credits refresh monthly Credits reset at the beginning of each calendar billing month. Unused credits do not roll over — your balance resets to your plan's full amount on your monthly renewal date. ## What happens when you run out of credits When your workspace exhausts its monthly credit allocation, Genloop pauses query processing. Members will see a message indicating that the workspace has reached its credit limit. No further queries run until either: * The monthly reset occurs and credits refresh, or * You upgrade to a higher plan with a larger credit allowance Running out of credits affects the entire workspace simultaneously — all members share the same pool. Monitor your usage to avoid disruption during critical periods. Liveboards continue to display their last computed results even after credits are exhausted. Only new query processing is paused. ## Check your credit usage Click your workspace name in the top-left corner and select **Settings**. Click **Billing** in the left sidebar. Click the **Usage** tab to see your current month's credit consumption broken down by member and by day. ## Upgrading mid-month When you upgrade your plan mid-month, Genloop immediately applies the new credit limit from your new plan. Your existing credit balance is replaced by the new plan's full monthly allocation — you don't receive a prorated top-up. For example: if you're on Free with 5 credits remaining and you upgrade to Pro mid-month, your balance becomes 100 credits immediately. If your team regularly approaches the credit limit before the month ends, upgrading a few days early gives you immediate access to your new allocation rather than waiting for the next billing cycle. ## Idle connectors and workspaces Idle connections and inactive workspaces do not consume credits. Credits are only spent when a query is actively processed. You can connect as many data sources as your plan allows, invite all your members, and build as many liveboards as you want — none of that affects your credit balance. ## Next steps Compare credit allocations across Free, Pro, Growth, and Enterprise. Manage seats and see who's using your workspace's credit pool. # Genloop Plans: Free, Pro, Growth, and Enterprise Source: https://docs.genloop.ai/plans/overview Compare Genloop's four plans — Free, Pro, Growth, and Enterprise — to find the right fit for your team size, data needs, and compliance requirements. Genloop offers four plans designed to grow with your team. Start for free with a single database connection, scale to multiple sources and larger teams on Pro and Growth, or bring the full platform — with custom deployment, SSO, and dedicated support — to your organization on Enterprise. Liveboards are always free on every plan. ## Plan comparison | Feature | Free | Pro | Growth | Enterprise | | --------------------------- | ------------------ | ------------------- | --------------------- | ------------------ | | **Price** | \$0/mo | \$20/mo | \$200/mo | Custom | | **Credits/mo** | 25 | 100 | 1,000 | Custom | | **DB connections** | 1 (up to 5 tables) | 3 (up to 20 tables) | 10 (up to 100 tables) | Unlimited | | **Members** | Up to 2 | Up to 20 | Up to 200 | Unlimited | | **Liveboards** | Free, always | Free, always | Free, always | Free, always | | **Slack integration** | Yes | Yes | Yes | Yes | | **MCP integration** | Yes | Yes | Yes | Yes | | **API access** | No | Yes | Yes | Yes | | **Automations** | No | Yes | Yes | Yes | | **Governance & RBAC** | No | No | Yes | Yes + custom roles | | **Role auto-sync (IdP)** | No | No | No | Yes | | **Build custom agents** | No | No | Yes | Yes | | **SSO / SAML** | No | No | No | Yes | | **Iframes / embedding** | No | No | No | Yes | | **VPC / air-gapped deploy** | No | No | No | Yes | | **Priority support** | No | No | Yes | Dedicated SA + SLA | ## Who each plan is for The Free plan is ideal for a solo analyst or a two-person team that wants to connect a single database and ask data questions without committing to a paid plan. You get 25 credits per month, Slack and MCP integrations, and access to liveboards at no cost. Pro is built for teams of up to 20 that are using Genloop regularly. You get 3 database connections, 100 credits per month, API access, and Automations — so you can trigger data queries programmatically and build lightweight workflows around your data. Growth is the right choice for organizations with 200 or fewer members that need to manage access controls, build custom agents, and scale across 10 data sources. Governance settings with RBAC let you define who sees what, and priority support means you get faster help when you need it. Enterprise is for organizations that need custom deployment (VPC or air-gapped), SSO and SAML, unlimited connections and users, custom roles with IdP sync, and a dedicated Solutions Architect. Pricing is tailored to your usage and deployment model. ## Upgrade your plan In the Genloop app, click your workspace name and select **Settings**. Click **Billing** in the left sidebar. Select the plan you want and click **Upgrade**. You'll be prompted to confirm payment details. When you upgrade mid-month, your new credit allowance and feature access apply immediately. See the [Credits](/plans/credits) page for details on what happens to your existing credit balance. ## Contact sales for Enterprise Enterprise pricing is custom and scoped to your organization's needs. To get a quote or learn more: * Book a call: [https://cal.com/ayush-genloop](https://cal.com/ayush-genloop) * Your dedicated Solutions Architect will guide you through deployment options, contract terms, and onboarding. ## Liveboards are always free Liveboards — collaborative dashboards that continuously update with live data — are included on every plan at no additional cost. You can create, share, and embed liveboards regardless of which plan you're on. Liveboards can be shared via public links on any paid plan. Iframe embedding for internal or external applications is available on Enterprise only. See the [Liveboards](/using-genloop/liveboards) page for details. ## Next steps See exactly what consumes a credit and how monthly resets work. Invite teammates and manage seats on your plan. # About Genloop Source: https://docs.genloop.ai/quickstart/about-genloop Learn what Genloop is, how its AI data intelligence platform delivers verified answers from your data warehouse, and why data teams trust it for analytics. Image Genloop is an enterprise data intelligence platform. It lets your team ask questions about your data in plain English and get verified, actionable answers. No SQL, no waiting on analysts, no dashboards to build. Most analytics tools give you a query result. Genloop gives you an answer with the reasoning behind it, the confidence to act on it, and a system that gets sharper every time you use it. ## What makes Genloop different Traditional BI tools require someone to know exactly what to ask and how to ask it. Genloop flips that. Anyone on your team can ask a question the way they would ask a colleague and get back an answer they can trust. Under the hood, three things make this possible: **Living Context Graph** When you connect your data, Genloop does not just map your schema. It builds a working model of your data: the relationships between tables, the logic behind your metrics, the business rules your team has established. This context is what separates a correct SQL query from an answer that actually makes sense for your business. **Deterministic Reasoning** Genloop does not guess. Every answer is traced back to your data, with the logic visible. If something looks off, you can see exactly where it came from and flag it. That feedback directly improves future answers. **Self-Learning Loop** Every question asked, every correction made, every piece of context added makes Genloop more accurate over time. It learns how your business uses its data, not just what the data looks like. ## What you can do with Genloop * **Ask your data anything** in plain English, from Chat. Follow up, drill down, investigate anomalies. * **Build Liveboards** to pin answers your team comes back to. They update automatically as your data does. * **Set up Automations** to schedule reports and stay on top of what matters. * **Connect everything**: databases, data files, business apps, unstructured documents. One place for all the context Genloop needs. ## Who uses Genloop Genloop is built for teams where data sits at the center of every decision and where the gap between having data and acting on it costs real time and money. Our customers include data-forward enterprises like Lenskart and NetApp. The people who use it daily range from heads of data and analytics leads to business teams who just need answers fast. ## How Genloop connects to your stack Genloop reads directly from your data sources. No copies, no pipelines to maintain. Connect a database, upload files, or bring in business apps your team already uses. Everything Genloop learns stays inside your environment. Genloop is SOC2 Type II and ISO 27001 certified. ## Next steps Set up your workspace and connect your data in about ten minutes. Explore Genloop in the preloaded demo workspace before connecting your own data. # Connect your data Source: https://docs.genloop.ai/quickstart/connect Connect your databases, files, and apps to Genloop so it can learn your tables, metrics, and business logic and return answers that reflect your business. Head to **Connections** in the left sidebar to get started. ## Two ways to connect ### Connect a database Supported sources: MySQL, PostgreSQL, Snowflake, BigQuery, Redshift, Databricks. This is a 4-step process: **Step 1 — Select your database** Choose your database type from the list of supported sources. Choose Database **Step 2 — Add credentials** Enter your connection details: host, port, database name, and credentials. Genloop connects directly to your source — no data copies are created. Credentials Credentials Credentials Credentials Make sure the credentials you're using have read access to the tables you want Genloop to work with. Write access is not required. **Step 3 — Select tables and columns** Choose which tables and columns you want Genloop to build memory on. You don't have to include everything — start with the data that matters most to your team's daily questions. At this step, you can also upload **business context documents** — things like metric definitions, KPI logic, data dictionaries, or business rules in `.md`, `.txt`, `.pdf`, or `.docx` format. This helps Genloop understand not just the shape of your data, but how your business actually uses it. The more context you add here, the better your answers get from day one. Select Tables Columns **Step 4 — Memory is created** Genloop processes your selected tables and builds your workspace context. This takes a few minutes. You'll see a progress stream as it runs. Reading Tables Columns Once complete, your workspace is live and you're ready to start asking questions. *** ### Upload a data file If you don't have a database connection ready, you can upload a data file directly. Supported formats: `.csv`, `.xlsx`, `.xls` At least one data file is required to move forward on this path. Data files are different from business context documents — data files are what Genloop builds memory on, while context documents (docs, markdown files) help Genloop understand your business rules and definitions. You can add both. Connect Data Once your file is uploaded and processed, your workspace is ready. *** ## Your workspace is set up Your data is connected and Genloop has built its initial context. From here, it keeps learning: every question you ask, every correction you give, every piece of feedback refines what it knows. ## Next steps Ask your first question in Chat and see how Genloop reasons through it. Review and edit what Genloop learned from your tables and documents. # Try Demo Data Source: https://docs.genloop.ai/quickstart/demo Try Genloop in the preloaded demo workspace, a permanent sandbox with sample data so you can test questions and reasoning before connecting your own. Every Genloop account comes with a demo workspace — a permanent sandbox with a preloaded dataset, ready the moment you sign in. It's separate from your own workspace and always available, so you can come back to it anytime. Use it to explore how Genloop reasons, test the investigation flow, or just get a feel for the platform before going deeper on your own data. ## Getting into the demo workspace From the workspace switcher in the top left, select **Demo Workspace**. You'll land directly in Chat with the demo dataset active — no connection setup, no credentials. Image Image ## Start exploring Try asking: * "What were total revenues last quarter?" * "Which product category had the highest return rate?" * "Show me the top customers by order value." Genloop returns an answer with its full reasoning visible — the tables it used, the logic it applied, and how it got to the number. Demo Workspace Chat The demo workspace is a good place to figure out what questions matter most to your team. Take note of what's useful — you can ask the same things once your own data is connected. ## What the demo workspace isn't The demo workspace is a sandbox — any context, feedback, or corrections you add here don't carry over to your own workspace. Think of it as a permanent playground, not a starting point that you graduate from. Your own workspace is where the real learning happens. Ready to bring in your actual data? Here's how. # First answer Source: https://docs.genloop.ai/quickstart/first-answer Ask your first question in Genloop Chat, review the AI's reasoning and SQL, validate the answer, and save it to a Liveboard to share with your team. ## Start in Chat **Chat** is your main workspace in Genloop. Type a question in plain English — Genloop figures out the rest. Start Chat Start with something you already know the answer to. A number you can verify. That way you're testing the reasoning, not just taking the output at face value. **Good first questions:** * "What were total sales last month?" * "Which region had the highest return rate in Q1?" * "Show me the top 10 customers by revenue this year." ## Read the reasoning Every answer comes with its reasoning visible. Genloop shows you the tables it queried, the logic it applied, and how it reached the number. Query Response This isn't just transparency for its own sake — it's how you catch something off and correct it early. If the logic looks wrong, flag it. That correction goes directly into improving future answers. ## Save what matters Once you have an answer worth keeping, pin it to a **Liveboard**. Liveboards are your team's shared view — living dashboards that update as your data does, built from the questions you actually asked. ## Where to go next Follow-ups, deep dives, anomaly investigation — everything Chat can do. Save answers and build shared views for your team. Add business definitions and metric logic to sharpen your answers. Bring in teammates and collaborate on the same data. # Quickstart Source: https://docs.genloop.ai/quickstart/overview Create your Genloop account, set up a workspace, and choose between exploring the demo dataset or connecting your own data, all in about ten minutes. Genloop connects to your data, builds a working understanding of your business, and lets you and your team ask questions in plain English — no SQL required, no dashboards to build, no tickets to raise. Getting started takes about 10 minutes. First, create your account at [app.genloop.ai](https://app.genloop.ai) using your work email. Personal emails aren't supported. Genloop uses your work domain to group your team's workspaces automatically. ## Create your workspace After signing up, Genloop asks you to name your workspace. This becomes your team's shared space — it holds your data connections, context, and chat history. Workspace Name ## Two workspaces, two purposes Once you're in, you'll have access to two separate workspaces: **Demo workspace** — preloaded and ready immediately. Every Genloop account comes with one. It's a permanent sandbox where you can explore the platform, test questions, and see how Genloop reasons — without touching your real data. **Your workspace** — the one you just named. This is where you connect your actual data and start getting answers that matter to your business. No setup needed. See Genloop in action on a preloaded dataset in under a minute. Connect a database or upload a file. Genloop builds context on your data and your workspace is live. # Compliance and Data Security at Genloop Source: https://docs.genloop.ai/security/compliance Genloop is SOC 2 Type II and ISO 27001 certified. Your data never leaves your infrastructure — Genloop reads directly from your sources. This page covers Genloop's security certifications, data handling practices, encryption standards, and audit capabilities. If you're evaluating Genloop for a regulated industry or enterprise deployment, use this page alongside your security questionnaire or vendor review process. ## SOC 2 Type II Genloop has completed a SOC 2 Type II audit, conducted by an independent third-party auditor. The audit covers the Trust Services Criteria for security, availability, and confidentiality — meaning Genloop's controls were tested over a period of time, not just at a single point. The SOC 2 Type II report covers: * Logical and physical access controls * Change management procedures * Risk assessment and monitoring * Incident response processes * Data encryption in transit and at rest The SOC 2 Type II report is available to customers and prospects under NDA. To request a copy, contact the Genloop team at [security@genloop.ai](mailto:security@genloop.ai) or reach out through your account manager. ## ISO 27001 Genloop is certified to ISO 27001, the international standard for information security management systems (ISMS). Certification means Genloop has implemented a comprehensive, risk-based framework for protecting information assets — covering policies, processes, personnel, and technology controls. ISO 27001 certification is reviewed annually by an accredited certification body. Contact [security@genloop.ai](mailto:security@genloop.ai) to request a copy of the ISO 27001 certificate. ## No data copies policy Genloop reads directly from your source tables at query time. It does not copy, replicate, or cache your enterprise data in its own storage layer. When you connect a database or data warehouse, Genloop: 1. Reads the schema to build a context graph 2. Issues queries against your source tables at runtime 3. Returns results to the user — nothing is stored Your data stays in your infrastructure. Genloop never moves it to a Genloop-controlled data store. Genloop does store conversation history and the context graph (schema metadata, business definitions, and learned context). See [Data retention](#data-retention) below for details on what you can delete. ## Encryption All data transmitted between your browser, the Genloop application, and your data sources is encrypted using TLS 1.2 or higher. This applies to API calls, Slack integration traffic, and the connection between Genloop and your database. Genloop does not store your source data, so there is no Genloop-side encryption requirement for your enterprise data. Data stored by Genloop — conversation history, context graph metadata, and workspace configuration — is encrypted at rest using AES-256. ## Audit logs Genloop logs the following events for all workspace activity: | Event category | What is logged | | ------------------- | -------------------------------------------------------------------------------- | | Queries | Query text, user, timestamp, data sources accessed, and result status | | User actions | Login events, role changes, member additions and removals | | Context updates | Every change to the context graph, including who proposed it and who approved it | | Data source changes | Connection added, removed, or modified; schema refreshes | | Liveboard actions | Created, edited, shared, or deleted | Audit logs are available to Admins in **Settings > Audit log**. Logs are retained for 90 days by default. Enterprise customers can request extended retention. Every context update Genloop learns goes through human-in-the-loop validation — your data team reviews and approves each change before it takes effect. These review actions are captured in the audit log. ## Data retention Genloop retains two categories of workspace data: **Conversation history** — the record of questions asked and answers returned within your workspace. This helps Genloop improve responses over time and lets your team review past analyses. **Context graph** — schema metadata, business definitions, column descriptions, and learned context about your data. This is the core knowledge layer that makes Genloop's answers accurate. You can delete either of these at any time: Navigate to **Settings > Data & privacy**. Select **Conversation history** to delete all past conversations, or **Context graph** to reset Genloop's learned knowledge about your data. Confirm the deletion. This action is permanent and cannot be undone. Deleting the context graph resets all learned business definitions and schema context. Genloop will re-learn context as your team uses it, but you may see reduced answer quality immediately after a reset. ## Requesting compliance documentation To request any compliance documentation — SOC 2 report, ISO 27001 certificate, data processing agreement (DPA), or security questionnaire responses — contact the Genloop security team: * Email: [security@genloop.ai](mailto:security@genloop.ai) * For enterprise procurement: contact your dedicated Solutions Architect or reach out at [https://cal.com/ayush-genloop](https://cal.com/ayush-genloop) ## Next steps Deploy Genloop in the cloud, in your VPC, or fully air-gapped. Govern access with built-in and custom roles. # Deployment Options and SSO Configuration Source: https://docs.genloop.ai/security/deployment Deploy Genloop in the cloud, inside your VPC, or fully air-gapped. Configure SSO and SAML for Enterprise workspaces with your identity provider. Genloop supports three deployment modes to meet different infrastructure and compliance requirements. You can start with the cloud deployment and migrate to a private deployment as your needs evolve. SSO and SAML authentication are available on the Enterprise plan for all deployment modes. ## Deployment modes **The default deployment.** Genloop is hosted and managed by the Genloop team. You connect your data sources, and Genloop handles infrastructure, updates, and scaling. **Best for:** Teams that want to get started immediately without infrastructure overhead. **Data handling:** Genloop queries your data sources directly. No enterprise data is copied or stored in Genloop's cloud. Connections use TLS-encrypted credentials stored in Genloop's secrets manager. No setup is required. Create your account at [app.genloop.ai](https://app.genloop.ai) and connect your first data source. **Genloop runs inside your cloud account.** The application is deployed into a Virtual Private Cloud (VPC) in your AWS, GCP, or Azure environment. Your data never leaves your cloud boundary. **Best for:** Organizations with strict data residency requirements or policies that prohibit third-party SaaS access to production databases. **Data handling:** All queries run within your VPC. Genloop's application layer, context graph, and conversation history are hosted in your environment. LLM API calls are made outbound from your VPC to the configured model provider. Reach out at [https://cal.com/ayush-genloop](https://cal.com/ayush-genloop) or contact your Solutions Architect to start the VPC deployment process. The Genloop team will provide infrastructure-as-code templates (Terraform or CloudFormation) and guide you through provisioning the required resources in your cloud account. Once deployed, configure your data source connections from within your VPC. Credentials remain inside your environment. **No external LLM calls.** All inference runs on-premises. The Genloop application and the underlying language model are deployed entirely within your environment — no traffic leaves your network. **Best for:** Government, defense, financial services, and other organizations with strict controls on external network egress or LLM data processing. **Data handling:** All data stays on-premises. You bring your own model (see [Model agnosticism](#model-agnosticism)). Air-gapped deployment requires Enterprise plan. Contact your Solutions Architect to discuss hardware requirements and deployment timelines. Air-gapped deployments are scoped individually. Reach out at [https://cal.com/ayush-genloop](https://cal.com/ayush-genloop) to begin the process. Choose the on-premises model you want Genloop to use. Genloop supports open-weight models compatible with standard inference runtimes. Your Solutions Architect will advise on tested and supported options. The Genloop team will provide deployment packages and work with your infrastructure team to install and configure the application within your environment. ## Model agnosticism Genloop is not tied to a single AI provider. Depending on your deployment mode, you can configure which underlying language model powers Genloop's analysis: * **Cloud (SaaS):** Genloop uses its default hosted model. Contact the team to discuss model options. * **VPC:** You can configure outbound LLM API calls to any supported provider (e.g., OpenAI, Anthropic, Azure OpenAI). * **Air-gapped:** You bring your own model. Genloop supports open-weight models served via standard inference APIs. Your Solutions Architect will confirm which runtimes and models are supported. Model selection does not affect Genloop's context graph, data access controls, or RBAC configuration. These are Genloop-layer features that work the same regardless of the underlying model. ## SSO and SAML SSO and SAML are available on the **Enterprise** plan for all deployment modes. Genloop supports SAML 2.0-based single sign-on, which lets your team log in with your existing identity provider instead of managing separate Genloop credentials. **Supported identity providers include:** * Okta * Microsoft Entra ID (formerly Azure AD) * Google Workspace * OneLogin * Any SAML 2.0-compliant provider ### Configure SSO SSO configuration is handled by the Genloop team during Enterprise onboarding. Contact your Solutions Architect or reach out at [https://cal.com/ayush-genloop](https://cal.com/ayush-genloop) to request activation. Share your identity provider's SAML metadata URL or XML file with the Genloop team. This includes your issuer URI, SSO URL, and public certificate. In your identity provider, create a new SAML application for Genloop. The Genloop team will provide the required **Entity ID**, **ACS URL** (Assertion Consumer Service), and attribute mapping requirements. Once configuration is complete, test the login flow with a non-admin account before rolling out to your full team. After a successful test, you can enforce SSO for all workspace members — disabling password-based login entirely. ### Role auto-sync with SSO Role auto-sync is available on **Enterprise** only. On Enterprise, Genloop can sync roles from your identity provider automatically. When a member's group membership changes in your IdP (for example, they're added to a `data-team` group), their Genloop role updates automatically at next login. This eliminates manual role management and ensures access controls stay in sync with your directory. Configure the group-to-role mapping in **Settings > Security > SSO** after your SSO integration is live. ## Requesting a private deployment To discuss VPC or air-gapped deployment: * Book a call with the team: [https://cal.com/ayush-genloop](https://cal.com/ayush-genloop) * Or contact your dedicated Solutions Architect if you're an existing Enterprise customer ## Next steps Review certifications, encryption standards, and audit logs. Set up the RBAC that pairs with SSO role auto-sync. # Agents Source: https://docs.genloop.ai/using-genloop/agents Explore Genloop's specialized AI agents for SQL generation, retrieval, forecasting, and root-cause analysis that work together to answer your data questions. Agents are the specialized AI capabilities Genloop uses under the hood to handle different types of questions. When you ask something in Chat, Genloop routes your question to the right agent depending on what you are asking and what data is involved. You do not configure agents to use them — they are already active and working every time you use Chat. The Agents page gives you visibility into which agents are available in your workspace. Agents ## What agents do Each agent is built for a specific job: * **Query Agent** answers factual questions about your data — counts, averages, distributions, and comparisons pulled directly from your connected tables. * **Analysis Agent** handles more complex requests that involve running Python code or multi-step reasoning across your data. * **Multimodal RAG Agent** reads your uploaded documents — PDFs, policy files, playbooks — and answers questions based on their content. * **Web Search Agent** fetches information from the web to answer questions that go beyond what is in your data. * **Report Agent** handles report generation and powers scheduled reports in Automations. ## Creating custom agents Custom agent creation is coming to the platform. You will be able to build and configure agents specific to your workflows directly from this page. Agents are workspace-level. Every member of your workspace has access to the same set of active agents. # Automations Source: https://docs.genloop.ai/using-genloop/automations Put any insight on a cadence, or a condition. Genloop stays silent until something matters, then delivers with root cause included, if you want it. Automations put any insight on a cadence: a time trigger for your recurring numbers, or a condition tied to a metric. Genloop watches continuously and stays silent until something matters. No digest noise, no daily emails you stop reading by week two. Every automation lives in its own chat, so you can open a row anytime to see its full run history and make changes. No separate builder to learn. Create one from scratch, or build it straight from a question in **Chat** or a tile on a **Liveboard**. Automations 2 ## The Automations list Every automation in your workspace shows up as a row: * **Name**: the name of the automation * **Source**: where the underlying question lives, **Chat** or **Liveboard**. Click through to open it. * **Notifies**: whether you hear about **every run** or **only when triggered** (see below) * **When it runs**: the cadence, for example *Daily, 9:00 AM* or *Weekly, Mon 8:00 AM* * **Last Run**: when the automation last ran * **Status**: whether the automation is active * **Action**: open the row's **⋮** menu to edit or delete it Use **Type to search…** to find an automation by name, or narrow the list with the **Source** and **Notifies** filters above the table. ## Creating an automation Click **Create New Automation** on the Automations page, or start from a question you're already asking: * From **Chat**, describe what you want in plain language: "Send me a weekly signups report every Monday at 9am" or "Alert me when daily orders drop more than 20% from the previous day." Genloop turns it into an automation, pre-filled and ready to review. * From a **Liveboard**, turn a tile into an automation the same way you'd create a report. Either way, you land on the same setup screen: Configure Automation * **Name**: what to call the automation * **What to ask**: the underlying question Genloop runs each time, in plain language * **Deep investigation**: an optional toggle that has Genloop surface anomalies and root cause alongside the result, whichever Notifies setting you use. Uses more credits per run. * **When to notify**: **Every run** sends the result each time, whether or not anything changed. **Only when triggered** stays silent until a condition you describe is true (for example, *"Weekly signups are above 10"*), then sends. * **When to run**: the cadence (daily, weekly, monthly, or custom), the days it checks, and the time. A line underneath confirms it in plain language, for example *"Checks weekly on Mon, Tue, Wed at 9:00 AM IST."* * **Send to**: the people who receive it, added by email The result reaches you where your team already works: in email, Slack, or Teams. Use **Test Run** to check the automation before committing, then **Activate Automation** to turn it on. An automation is only as good as the question behind it. If you ask for something Genloop cannot compute from your connected data, tighten the question in **What to ask** before activating. ## Next steps Build the Liveboards you can turn into automations. Turn a scheduled automation into a full report or presentation. # Chat Source: https://docs.genloop.ai/using-genloop/chat Ask data questions in plain English with Genloop Chat. Get verified answers with a transparent query plan, editable SQL, charts, and forecasts of what happens next, with multi-step agentic analysis when a question needs it. Chat is where you spend most of your time in Genloop. Type a question in plain English, and Genloop returns a verified answer with the full reasoning behind it — the tables it used, the logic it applied, and how it reached the result. Think of it less like a search box and more like working with an analyst. You can ask a quick question, see exactly how Genloop reached the answer, adjust the SQL if you want, and follow up until you have what you need — all in one conversation. Start a chat in Genloop ## Starting a chat Click **New Chat** in the sidebar to open a fresh thread. Genloop greets you with a few **suggested questions** based on your connected data — click one to run it instantly, or write your own in the input bar at the bottom. The chips above the input — **Lookup**, **Compare**, **Trend**, **Why**, and **Deep Dive** — are question starters. Selecting one surfaces example questions of that kind to help you phrase your ask. They are prompts to get you going, not settings; Genloop decides how to answer based on the question itself, not the chip you picked. Every thread stays in the sidebar under **Chats**, so you can pick up an earlier conversation at any time or start over with **New Chat**. ## Asking a question Type your question the way you would ask a colleague. You can be as **specific** or as **exploratory** as you want: * **Specific** — *"What is the average credit score in the latest financial health snapshot?"* * **Exploratory** — *"Which loan types are driving the most delinquency, and why?"* Two more ways to enter a question: * **Attach a file** — Click the paperclip to add a file for Genloop to reference alongside your connected data. * **Voice input** — Click the microphone to dictate your question instead of typing. Treat Genloop like a coworker. Have a back-and-forth — one question at a time gets sharper answers than one long, multi-part prompt. ## How Genloop answers Genloop reads every question and decides, on its own, how much work it needs. Simple lookups run as a single query. Anything that needs several steps — comparisons, root-cause analysis, documents, the web, or a report — is handled agentically. You do not choose the path; Genloop routes each question at runtime based on what it is actually asking. A direct, single-step lookup answerable with **one SQL query against one source**. You get the generated **SQL**, the **result** table, and a **chart** — right away, no extra steps. A question that needs **multiple steps**. Genloop builds an **execution plan**, may pause to **clarify or get your approval**, runs the steps across specialized agents, and **synthesizes** the results into one answer. ### Simple questions For a straightforward lookup, Genloop writes a single SQL query, runs it, and returns the answer as a written summary plus a table and chart. Expand the **Query Plan** above the answer to see the exact query it ran — for example, a `SELECT` with a `GROUP BY` and `ORDER BY`. This is not just for transparency; it is how you catch a wrong assumption before it becomes a decision. A Genloop Chat answer with its Query Plan, summary, and results table ### Agentic questions When a question needs more than one step, Genloop plans the work as an **execution plan** — a sequence of steps, each handled by the right specialized [agent](/using-genloop/agents) (querying a database, searching your documents, running Python analysis, searching the web, or compiling a report). Steps that do not depend on each other run in parallel, and the results are synthesized into a single answer. An agentic execution plan with numbered steps and an approval prompt Two things can happen before or during an agentic run: * **Clarifying questions** — If your question is ambiguous, Genloop pauses and asks a follow-up before it runs anything — for example, *"Which date range?"* or *"Which customer segment?"* You answer inline, and Genloop folds your answer into the question and continues. These prompts are blocking: the run waits on your response. * **Plan approval** — For a substantial multi-step plan, Genloop can show you the plan and wait for your go-ahead before executing. Reply to approve it, or ask for a change and Genloop revises the plan and checks again. Simple, single-step questions skip this and run straight through. The query plan and the approval step exist so you stay in control. Genloop can make mistakes — reviewing the plan before it runs is your chance to steer it. ### Asking about the future Most questions in Chat look backward: last quarter, last month, last week. Some don't. Ask "which category will drive revenue next quarter" or "which customers are likely to churn," and Genloop builds a predictive model on your data and returns the projection the same way it returns any other answer: with the reasoning visible, on your governed data. There is no separate ML workflow to set up and no handoff to a data science queue. It is the same Chat, the same Query Plan, the same investigate-and-refine loop, just pointed at what happens next instead of what already happened. ## Reviewing and editing the query Every answer keeps its query available for inspection. Open the **Query Plan** to read the steps, and open **SQL Query** to **view or edit** the generated SQL directly. If you want to adjust the logic — tweak a filter, change a grouping, fix a join — edit the SQL in place and click **Run** to execute your version. Genloop returns an updated answer from your edited query, so you can refine the result without rephrasing the whole question. Editing a tile's SQL and re-running it from the Query tab ## Reading the answer Below the written summary, the result is shown as a table or a chart. Switch between **Table** and **Visualisation** with the toggle above the result. In **table** view you can: * **Search** across all rows * Click a column header to **sort** (ascending, descending, or off) * Use **Columns** to choose which columns are shown, and filter a column by value or condition * Expand with **Show more rows** when a result is long In **chart** view, Genloop answers with the chart type that fits the question: bar, line, funnel, heat map, stacked column, area, and more. If you'd rather see it differently, switch chart types directly on the answer, and adjust the metric, the aggregation (sum, average, count, min, max), and dimension filters to reshape the visualization. Use **Download** to export: the chart as an image, or the data as CSV (complete or filtered). ### When one chart isn't enough Some questions need more than a single chart to answer well. Ask something like "how did revenue move across channels and categories this quarter," and Genloop weaves multiple charts into one composed answer instead of making you ask three follow-up questions. Every chart in that answer keeps the same controls: switch its type, adjust its axes, or change its breakdown on its own. Switch Chart ## Digging deeper After an answer, Genloop suggests where to go next under **Explore Further** — one-click follow-up questions that build on what you just saw. Unlike clarifying questions, these are optional; click one to send it, or keep typing your own. Because follow-ups happen in the same thread, the conversation builds naturally on everything that came before. ## What you can do with an answer Every answer is a starting point, not an endpoint. Each one carries a set of actions: Pin the answer as a tile on a [Liveboard](/using-genloop/liveboards), choosing chart or table view first. The tile stays connected to your live data and refreshes automatically. Save something worth keeping as business context. Genloop asks you to confirm the answer is correct, then proposes it as a memory update for your team to approve. Confirm an answer is right, or flag it as wrong and add a note on what went wrong. This feedback is aggregated for your data team. Send the answer to a data expert to review in the [Review Center](/using-genloop/review-center), where they can approve the logic as a trusted, verified result. Together, these actions are how Chat gets smarter over time — every rating, validation, and remembered fact feeds Genloop's continuous learning about how your business works. ## Tips for better results * **Start with questions you already know the answer to.** Verify the reasoning before you rely on it for decisions you are less certain about. * **Break complex asks into steps.** One question at a time gets better results than a single sprawling prompt. * **Read the Query Plan when a number looks off.** The fastest way to spot a wrong join or filter is to look at the SQL — and you can edit and re-run it if needed. * **Teach it as you go.** When Genloop gets something right that it did not know before, use **Remember This** so it applies that context next time. ## Next steps Pin answers into a shared dashboard that updates as your data does. See which specialized agents handle your questions behind the scenes. # Connect a database Source: https://docs.genloop.ai/using-genloop/connections/connect-a-database Connect a database to Genloop: grant network access, enter credentials or an SSH tunnel, select tables, and generate memory your team can query in natural language. Genloop connects to your database with read-only access and builds queryable memory from the tables you select. It reads from your source directly, without copying or moving your dataGenloop connects to your database with read-only access and builds queryable memory from the tables you select. It reads from your source directly, without copying or moving your data. Supported engines: PostgreSQL, MySQL, Snowflake, BigQuery, Redshift, Oracle, Aurora, ClickHouse, Databricks, Trino, Azure SQL, and SAP HANAengines: PostgreSQL, MySQL, Snowflake, BigQuery, Redshift, Oracle, Aurora, ClickHouse, Databricks, Trino, Azure SQL, and SAP HANA. Open the **Structured Data** tab and click **Connect database** to beginOpen the **Structured Data** tab and click **Connect database** to begin. Structured Data connections list ## Before you begin The **Add New Connection** screen opens with a **Before you start** panel. Both steps below are optional. Complete them only if they apply to you. Before you start panel with invite and IP allowlist **If you do not have the credentials**, enter a colleague's email under **Get credentials** and click **Send Invite**. They can open the connection and complete the setup for you. Structured Data connections list ## Before you begin The **Add New Connection** screen opens with a **Before you start** panel. Both steps below are optional. Complete them only if they apply to you. Before you start panel with invite and IP allowlist **If you do not have the credentials**, enter a colleague's email under **Get credentials** and click **Send Invite**. They can open the connection and complete the setup for you. **If your database is behind a firewall**, click **Copy IP address**, select your region, and add the address to your database's inbound rules. Skip this step if your database already accepts connections from any IP addres**If your database is behind a firewall**, click **Copy IP address**, select your region, and add the address to your database's inbound rules. Skip this step if your database already accepts connections from any IP address. | Region | IP address | | ------------ | -------------- | | `us-east-1` | `52.21.2.252` | | `ap-south-1` | `13.233.59.14` | ## Connect your database Select your engine from the supported list, then click **Add connection details**. Choose a database type Name the connection, confirm the **Data Source**, and complete the fields for your database. Enter host, port, database, and credentials | Field | Description | | ------------ | -------------------------------------------------------------- | | **Host** | Hostname or IP address of your database server | | **Port** | Port your database listens on. Each engine has its own default | | **Database** | Name of the database to connect | | **Username** | User Genloop connects as | | **Password** | Password for that user | The connecting user needs read access only. Genloop never writes to your source. Field names and defaults vary by engine — see the [database guides](/using-genloop/connections/databases/mysql) for yours. If you reach the database through a bastion or jump host, enable **Connect via SSH tunnel** and complete the tunnel fields. SSH tunnel fields including host, username, and PEM key | Field | Description | | ----------------------- | ------------------------------------------------------------------------------------------- | | **SSH host** | Public IP or DNS of the bastion. Cannot be `localhost` | | **SSH port** | SSH port on the bastion. Default is `22` | | **SSH username** | User Genloop signs in to the bastion as | | **PEM key** | Your private key file. Sent encrypted with the rest of the credentials | | **Remote bind address** | Database host as seen from the bastion. Use `127.0.0.1` if the database runs on the bastion | | **Remote bind port** | Database port as reached from the bastion | Click **Connect & Save**. Genloop validates the connection and opens table selection when it succeeds. A failed validation names the exact problem to fix, usually a credential, a port, or an IP address that still needs to be allowed. Choose the tables and columns to build memory on. Select a table to view its columns, and use **Select all** or **Clear** to adjust. Start with the tables that answer your team's most common questions; you can add more later. The footer totals your selected **Tables**, **Columns**, and **Workspace Tables** against your plan limit. Select tables and columns to onboard Click **Generate memory**. Genloop builds the semantic context: column descriptions, table relationships, and business meaning. Progress appears as each stage completes. When it finishes, the connection is live and ready to query in Chat. ## Keep tables in sync Schemas change as tables and columns are added or dropped. Open a connection and click **Refresh** to re-read the live schema. Genloop detects the drift and shows what changed, so your selection stays aligned with the database. ## Add more tables later Onboard in stages. Tables that already have memory show an **In Context** badge. To add tables you skipped, reopen the connection, select the remaining tables or columns, and click **Generate memory**. Genloop processes only the new selection and leaves your existing memory unchanged. Delta onboarding — already-onboarded tables marked In Context ## Remove a connection Connections appear in the **Structured Data** tab with their source, last successful sync, and latest sync time. Open the **⋯** menu on a row to manage it. Removing a connection is permanent. It deletes the connection and everything derived from it, including generated semantics, stored credentials, and its registry entry. To pick up schema changes without losing memory, use **Refresh** instead. Database Type Remove connection confirmation Select Tables And Columns Business Context | Region | IP address | | ------------ | -------------- | | `us-east-1` | `52.21.2.252` | | `ap-south-1` | `13.233.59.14` | ## Connect your database Select your engine from the supported list, then click **Add connection details**. Choose a database type Name the connection, confirm the **Data Source**, and complete the fields for your database. Enter host, port, database, and credentials | Field | Description | | ------------ | -------------------------------------------------------------- | | **Host** | Hostname or IP address of your database server | | **Port** | Port your database listens on. Each engine has its own default | | **Database** | Name of the database to connect | | **Username** | User Genloop connects as | | **Password** | Password for that user | The connecting user needs read access only. Genloop never writes to your source. Field names and defaults vary by engine — see the [database guides](/using-genloop/connections/databases/mysql) for yours. If you reach the database through a bastion or jump host, enable **Connect via SSH tunnel** and complete the tunnel fields. SSH tunnel fields including host, username, and PEM key | Field | Description | | ----------------------- | ------------------------------------------------------------------------------------------- | | **SSH host** | Public IP or DNS of the bastion. Cannot be `localhost` | | **SSH port** | SSH port on the bastion. Default is `22` | | **SSH username** | User Genloop signs in to the bastion as | | **PEM key** | Your private key file. Sent encrypted with the rest of the credentials | | **Remote bind address** | Database host as seen from the bastion. Use `127.0.0.1` if the database runs on the bastion | | **Remote bind port** | Database port as reached from the bastion | Click **Connect & Save**. Genloop validates the connection and opens table selection when it succeeds. A failed validation names the exact problem to fix, usually a credential, a port, or an IP address that still needs to be allowed. Choose the tables and columns to build memory on. Select a table to view its columns, and use **Select all** or **Clear** to adjust. Start with the tables that answer your team's most common questions; you can add more later. The footer totals your selected **Tables**, **Columns**, and **Workspace Tables** against your plan limit. Select tables and columns to onboard Click **Generate memory**. Genloop builds the semantic context: column descriptions, table relationships, and business meaning. Progress appears as each stage completes. When it finishes, the connection is live and ready to query in Chat. ## Keep tables in sync Schemas change as tables and columns are added or dropped. Open a connection and click **Refresh** to re-read the live schema. Genloop detects the drift and shows what changed, so your selection stays aligned with the database. ## Add more tables later Onboard in stages. Tables that already have memory show an **In Context** badge. To add tables you skipped, reopen the connection, select the remaining tables or columns, and click **Generate memory**. Genloop processes only the new selection and leaves your existing memory unchanged. Delta onboarding — already-onboarded tables marked In Context ## Remove a connection Connections appear in the **Structured Data** tab with their source, last successful sync, and latest sync time. Open the **⋯** menu on a row to manage it. Removing a connection is permanent. It deletes the connection and everything derived from it, including generated semantics, stored credentials, and its registry entry. To pick up schema changes without losing memory, use **Refresh** instead. Database Type Remove connection confirmation Select Tables And Columns Business Context ## Next steps Field-by-field setup for your specific databaseField-by-field setup for your specific database. Add CSV or Excel files alongside your databaseAdd CSV or Excel files alongside your database. View and edit the memory Genloop built from your connection. Add business documents that inform Genloop's answers. # Connect an app Source: https://docs.genloop.ai/using-genloop/connections/connect-an-app Connect third-party apps like Salesforce, Google Analytics, and HubSpot to Genloop with OAuth and sync their data alongside your warehouse to query together. Genloop connects to the tools your team already uses — bringing in that data as additional context alongside your databases and files. Go to **Connections** and click the **Apps** tab to see the full list of supported integrations. \[Screenshot: Apps tab showing integration tiles] ## Connecting an app Click on any app to start the connection flow. Genloop will walk you through the authorization steps specific to that tool. Once connected, the app appears in your **Connected Apps** panel on the right. ## Supported apps Google Analytics, Salesforce, Slack, HubSpot, Jira, GitHub, Notion, Asana, Zendesk, Airtable, Intercom, Confluence, Dropbox, Google Drive, OneDrive, and more. Don't see an integration you need? [Talk to us](https://cal.com/ayush-genloop) and let us know what you would like connected. ## Next steps Add structured database connections alongside your app integrations. See how connected app data feeds into your workspace context. # Aurora Source: https://docs.genloop.ai/using-genloop/connections/databases/aurora Connect an Amazon Aurora cluster to Genloop in MySQL or PostgreSQL mode with cluster endpoint and read-only credentials, optionally through an SSH tunnel. Genloop connects to Amazon Aurora over a direct connection using a read-only user. Aurora runs in one of two modes, MySQL or PostgreSQL, and you tell Genloop which one your cluster uses. ## Before you connect * An Aurora user with read access to the tables you want to query. * Whether your cluster is **MySQL** or **PostgreSQL** compatible. * Network access from Genloop. If your cluster is in a VPC, allow Genloop's regional IP address in its security group. See [Before you begin](/using-genloop/connections/connect-a-database#before-you-begin). ## Connection fields | Field | Description | | ------------------- | --------------------------------------------------------------------------------------------- | | **Connection name** | A name to identify this connection in Genloop | | **Engine** | The cluster's compatibility mode: `mysql` or `postgresql` | | **Host** | Your Aurora cluster endpoint | | **Port** | The port your cluster listens on. Defaults to `3306` (MySQL mode) or `5432` (PostgreSQL mode) | | **Database** | Name of the database to connect | | **Username** | Aurora user Genloop connects as | | **Password** | Password for that user | ## Connect through an SSH tunnel If the cluster is not reachable directly, reach it through a bastion or jump host. Turn on **Connect via SSH tunnel** in the connection form and complete these fields. | Field | Description | | ----------------------- | ----------------------------------------------------------------------------------------- | | **SSH host** | Public IP or DNS of the bastion. Cannot be `localhost` | | **SSH port** | SSH port on the bastion. Default is `22` | | **SSH username** | User Genloop signs in to the bastion as | | **PEM key** | Your private key file. Sent encrypted with the rest of the credentials | | **Remote bind address** | Cluster host as seen from the bastion. Use `127.0.0.1` if the cluster runs on the bastion | | **Remote bind port** | Cluster port as reached from the bastion | ## Permissions Genloop only reads your data and never writes to your cluster. You can connect with any user that already has read access to the tables you select. If you want a dedicated read-only user for Genloop, ask your database administrator to set one up. Run the commands for your cluster's engine: ```sql theme={null} -- MySQL-compatible GRANT SELECT ON your_database.* TO 'genloop_user'@'%'; -- PostgreSQL-compatible GRANT USAGE ON SCHEMA public TO genloop_user; GRANT SELECT ON ALL TABLES IN SCHEMA public TO genloop_user; ``` ## Next steps Back to the full connection flow. View and edit the memory built from your connection. # Azure SQL Source: https://docs.genloop.ai/using-genloop/connections/databases/azure-sql Connect an Azure SQL database to Genloop with server, database, and read-only credentials over an encrypted connection, optionally through an SSH tunnel. Genloop connects to Azure SQL over an encrypted connection using a read-only user. The same connector works for self-hosted SQL Server. ## Before you connect * A database user with read access to the tables you want to query. * Network access from Genloop. In the Azure portal, add Genloop's regional IP address to your server's firewall rules. See [Before you begin](/using-genloop/connections/connect-a-database#before-you-begin). ## Connection fields | Field | Description | | ---------------------------- | ------------------------------------------------------------------------------------- | | **Connection name** | A name to identify this connection in Genloop | | **Host** | Server hostname, for example `myserver.database.windows.net` | | **Port** | The port your server listens on. Defaults to `1433` | | **Database** | Name of the database to connect | | **Username** | Database user Genloop connects as | | **Password** | Password for that user | | **Trust server certificate** | Off by default. Enable only for self-hosted SQL Server with a self-signed certificate | ## Encryption (TLS) The connection is always encrypted; there is no option to turn encryption off. What you control is certificate validation: * **Azure SQL** — leave **Trust server certificate** off so the server's certificate is validated. * **Self-hosted SQL Server with a self-signed certificate, or a tunnelled connection to a local server** — turn **Trust server certificate** on so the connection is accepted. ## Connect through an SSH tunnel If the server is not reachable directly, reach it through a bastion or jump host. Turn on **Connect via SSH tunnel** in the connection form and complete these fields. When tunnelling to a local server, turn **Trust server certificate** on. | Field | Description | | ----------------------- | --------------------------------------------------------------------------------------- | | **SSH host** | Public IP or DNS of the bastion. Cannot be `localhost` | | **SSH port** | SSH port on the bastion. Default is `22` | | **SSH username** | User Genloop signs in to the bastion as | | **PEM key** | Your private key file. Sent encrypted with the rest of the credentials | | **Remote bind address** | Server host as seen from the bastion. Use `127.0.0.1` if the server runs on the bastion | | **Remote bind port** | Server port as reached from the bastion | ## Permissions Genloop only reads your data. Give the connecting user read access to the tables you select; no write access is needed. A minimal grant looks like this: ```sql theme={null} GRANT SELECT ON SCHEMA::dbo TO genloop_user; ``` ## Next steps Back to the full connection flow. View and edit the memory built from your connection. # BigQuery Source: https://docs.genloop.ai/using-genloop/connections/databases/bigquery Connect a Google BigQuery project to Genloop with a service account key, scope access to a dataset, and query your warehouse in natural language. Genloop connects to BigQuery over an encrypted (HTTPS) connection using a Google Cloud service account. You provide the service account key and the dataset you want Genloop to read. ## Before you connect * A Google Cloud service account with read access to your data. * The dataset you want Genloop to query. ## Connection fields | Field | Description | | ------------------------------ | ------------------------------------------------------------------------------------ | | **Connection name** | A name to identify this connection in Genloop | | **Service account key (JSON)** | The JSON key file for your service account | | **Dataset** | The dataset Genloop reads. May be given as `project.dataset` | | **Project** | Optional. Project where the dataset lives. Defaults to the service account's project | | **Location** | Optional. Region where queries run | ## Set up a service account In the Google Cloud Console, go to **IAM & Admin → Service Accounts**. Create or select a service account and grant it **BigQuery Data Viewer** and **BigQuery Job User**. Create a key for the service account, download the JSON file, and upload it in the Genloop connection form. ## Encryption and network Genloop reaches BigQuery over the internet using an encrypted (HTTPS) connection. There is no SSL setting to configure, and an SSH tunnel is not used. Control access through the service account's roles rather than a network allowlist. ## Permissions Genloop only reads your data. The service account you provide needs read access to the datasets you want to query. If you want a dedicated service account for Genloop, ask your Google Cloud administrator to grant it the roles below. Grant the service account: * `bigquery.dataViewer` on the datasets you want Genloop to query. * `bigquery.jobUser` on the project to run queries. ## Next steps Back to the full connection flow. View and edit the memory built from your connection. # ClickHouse Source: https://docs.genloop.ai/using-genloop/connections/databases/clickhouse Connect a ClickHouse database to Genloop over HTTP or HTTPS with host, port, and read-only credentials, optionally through an SSH tunnel. Genloop connects to ClickHouse over its HTTP interface using a read-only user. By default the connection is encrypted (HTTPS). ## Before you connect * A ClickHouse user with read access to the tables you want to query. * Whether the server uses a secure (HTTPS) or plain (HTTP) interface. * Network access from Genloop. If the server is behind a firewall, allow Genloop's regional IP address. See [Before you begin](/using-genloop/connections/connect-a-database#before-you-begin). ## Connection fields | Field | Description | | ------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Connection name** | A name to identify this connection in Genloop | | **Host** | Hostname or IP address of your ClickHouse server | | **Port** | Port of your ClickHouse HTTP interface. Defaults to `8443`. Commonly `8443` for HTTPS or `8123` for HTTP, but use whatever your server is configured for | | **Database** | Name of the database to query | | **Username** | ClickHouse user Genloop connects as | | **Password** | Password for that user | | **Secure** | Use HTTPS. Enabled by default. Disable only for a plain HTTP endpoint | ## Encryption (TLS) Keep **Secure** turned on to connect over HTTPS. This is the recommended setting and keeps traffic encrypted. Turn it off only when your server's HTTP interface is not encrypted, for example inside a trusted private network. Set **Port** to match the interface you use. The port does not change automatically when you toggle **Secure**, so make sure it points at your server's actual HTTP or HTTPS port. ## Connect through an SSH tunnel If the server is not reachable directly, reach it through a bastion or jump host. Turn on **Connect via SSH tunnel** in the connection form and complete these fields. | Field | Description | | ----------------------- | --------------------------------------------------------------------------------------- | | **SSH host** | Public IP or DNS of the bastion. Cannot be `localhost` | | **SSH port** | SSH port on the bastion. Default is `22` | | **SSH username** | User Genloop signs in to the bastion as | | **PEM key** | Your private key file. Sent encrypted with the rest of the credentials | | **Remote bind address** | Server host as seen from the bastion. Use `127.0.0.1` if the server runs on the bastion | | **Remote bind port** | Server port as reached from the bastion | ## Permissions Genloop only reads your data and never writes to your database. You can connect with any user that already has read access to the tables you select. If you want a dedicated read-only user for Genloop, ask your database administrator to set one up. Run this in your ClickHouse database: ```sql theme={null} GRANT SELECT ON your_database.* TO genloop_user; ``` ## Next steps Back to the full connection flow. View and edit the memory built from your connection. # Databricks Source: https://docs.genloop.ai/using-genloop/connections/databases/databricks Connect a Databricks SQL warehouse to Genloop with a personal access token, scope access to a Unity Catalog catalog, and query your lakehouse in natural language. Genloop connects to a Databricks SQL warehouse over an encrypted (HTTPS) connection using a personal access token. It reads the tables you select and never writes to your workspace. ## Before you connect * A Databricks user or service principal with read access to the tables you want to query. * A running SQL warehouse and its HTTP path. * If your workspace has an IP access list, allow Genloop's regional IP address. See [Before you begin](/using-genloop/connections/connect-a-database#before-you-begin). ## Connection fields | Field | Description | | ------------------- | -------------------------------------------------------------------------------------- | | **Connection name** | A name to identify this connection in Genloop | | **Server hostname** | Your Databricks workspace hostname, for example `adb-1234567890.1.azuredatabricks.net` | | **HTTP path** | HTTP path of your SQL warehouse, found under its connection details | | **Access token** | A personal access token for authentication | | **Catalog** | Optional. Unity Catalog catalog to use. Defaults to the workspace default | ## Generate an access token In your Databricks workspace, click your username and open **User Settings → Developer**. Click **Generate new token**, set an expiry, and copy the value. Paste the token into the **Access token** field in the connection form. ## Encryption and network Genloop reaches Databricks over the internet using an encrypted (HTTPS) connection. There is no SSL setting to configure, and an SSH tunnel is not used. ## Permissions Genloop only reads your data and never writes to your workspace. You can connect with any user or service principal that already has read access to the tables you select. If you want a dedicated read-only identity for Genloop, ask your Databricks administrator to set one up. Grant the identity `CAN USE` on the SQL warehouse, then run this for the data it should read: ```sql theme={null} GRANT USE CATALOG ON CATALOG your_catalog TO `genloop_user`; GRANT USE SCHEMA ON SCHEMA your_catalog.your_schema TO `genloop_user`; GRANT SELECT ON TABLE your_catalog.your_schema.your_table TO `genloop_user`; ``` ## Next steps Back to the full connection flow. View and edit the memory built from your connection. # MySQL Source: https://docs.genloop.ai/using-genloop/connections/databases/mysql Connect a MySQL database to Genloop with host, port, and read-only credentials, optionally through an SSH tunnel, and query your tables in natural language. Genloop connects to MySQL over a direct database connection using a read-only user. It reads your schema and the tables you select, and never writes to your database. ## Before you connect * A MySQL user with read access to the tables you want to query. * Network access from Genloop. If your database is behind a firewall, allow Genloop's regional IP address. See [Before you begin](/using-genloop/connections/connect-a-database#before-you-begin). Genloop connects to MySQL over a direct database connection using a read-only user. It reads your schema and the tables you select, and never writes to your database. ## Before you connect * A MySQL user with read access to the tables you want to query. * Network access from Genloop. If your database is behind a firewall, allow Genloop's regional IP address. See [Before you begin](/using-genloop/connections/connect-a-database#before-you-begin). ## ConnectionConnection fields | Field | Description | | ------------------- | ----------------------------------------------------------------------------------------------------------- | | **Connection name** | A name to identify this connection in Genloop | | **Host** | Hostname or IP address of your MySQL serverHostname or IP address of your MySQL server | | **Port** | The port your MySQL server listens on. Defaults toThe port your MySQL server listens on. Defaults to `3306` | | **Database** | Name of the databaseName of the database to connect | | **Username** | MySQL user Genloop connects asMySQL user Genloop connects as | | **Password** | Password for thatPassword for that user | ## Connect through an SSH tunnelConnect through an SSH tunnel If the database is not reachable directly, reach it through a bastion or jump host. Turn on **Connect via SSH tunnel** in the connection form and complete these fieldsIf the database is not reachable directly, reach it through a bastion or jump host. Turn on **Connect via SSH tunnel** in the connection form and complete these fields. | Field | Description | | ----------------------- | ------------------------------------------------------------------------------------------- | | **SSH host** | Public IP or DNS of the bastion. Cannot be `localhost` | | **SSH port** | SSH port on the bastion. Default is `22` | | **SSH username** | User Genloop signs in to the bastion as | | **PEM key** | Your private key file. Sent encrypted with the rest of the credentials | | **Remote bind address** | Database host as seen from the bastion. Use `127.0.0.1` if the database runs on the bastion | | **Remote bind port** | Database port as reached from the bastion | | Field | Description | | ----------------------- | ------------------------------------------------------------------------------------------- | | **SSH host** | Public IP or DNS of the bastion. Cannot be `localhost` | | **SSH port** | SSH port on the bastion. Default is `22` | | **SSH username** | User Genloop signs in to the bastion as | | **PEM key** | Your private key file. Sent encrypted with the rest of the credentials | | **Remote bind address** | Database host as seen from the bastion. Use `127.0.0.1` if the database runs on the bastion | | **Remote bind port** | Database port as reached from the bastion | ## PermissionsPermissions Genloop only reads your data and never writes to your database. You can connect with any user that already has read access to the tables you selectGenloop only reads your data and never writes to your database. You can connect with any user that already has read access to the tables you select. If you want a dedicated read-only user for Genloop, ask your database administrator to set one up. Run this in your MySQL database: ```sql theme={null} GRANT SELECT ON your_database.* TO 'genloop_user'@'%'; ``` If you want a dedicated read-only user for Genloop, ask your database administrator to set one up. Run this in your MySQL database: ```sql theme={null} GRANT SELECT ON your_database.* TO 'genloop_user'@'%'; ``` ## Next steps Back to the full connection flow. View and edit the memory built from your connection. # Oracle Source: https://docs.genloop.ai/using-genloop/connections/databases/oracle Connect an Oracle database to Genloop with host, port, a service name or SID, and read-only credentials, optionally through an SSH tunnel. Genloop connects to Oracle over a direct database connection using a read-only user. You identify the target instance by either a service name or a SID. ## Before you connect * An Oracle user with read access to the tables you want to query. * The instance's **service name** (modern, multitenant databases) or **SID** (legacy, single-instance databases). Provide one. * Network access from Genloop. If the database is behind a firewall, allow Genloop's regional IP address. See [Before you begin](/using-genloop/connections/connect-a-database#before-you-begin). ## Connection fields | Field | Description | | ------------------- | ----------------------------------------------------------------------- | | **Connection name** | A name to identify this connection in Genloop | | **Host** | Hostname or IP address of your Oracle server | | **Port** | The port your Oracle server listens on. Defaults to `1521` | | **Username** | Oracle user Genloop connects as | | **Password** | Password for that user | | **Service name** | Service name of the target instance. Provide this or a SID | | **SID** | System identifier for a legacy instance. Provide this or a service name | ## Connect through an SSH tunnel If the database is not reachable directly, reach it through a bastion or jump host. Turn on **Connect via SSH tunnel** in the connection form and complete these fields. | Field | Description | | ----------------------- | ------------------------------------------------------------------------------------------- | | **SSH host** | Public IP or DNS of the bastion. Cannot be `localhost` | | **SSH port** | SSH port on the bastion. Default is `22` | | **SSH username** | User Genloop signs in to the bastion as | | **PEM key** | Your private key file. Sent encrypted with the rest of the credentials | | **Remote bind address** | Database host as seen from the bastion. Use `127.0.0.1` if the database runs on the bastion | | **Remote bind port** | Database port as reached from the bastion | ## Permissions Genloop only reads your data and never writes to your database. You can connect with any user that already has permission to sign in and read the tables you select. If you want a dedicated read-only user for Genloop, ask your database administrator to set one up. Run this in your Oracle database: ```sql theme={null} GRANT CREATE SESSION TO genloop_user; GRANT SELECT ON your_schema.your_table TO genloop_user; ``` ## Next steps Back to the full connection flow. View and edit the memory built from your connection. # PostgreSQL Source: https://docs.genloop.ai/using-genloop/connections/databases/postgresql Connect a PostgreSQL database to Genloop with host, port, and read-only credentials, optionally through an SSH tunnel, and query your schemas in natural language. Genloop connects to PostgreSQL over a direct database connection using a read-only user. It reads your schema and the tables you select, and never writes to your database. ## Before you connect * A PostgreSQL user with read access to the tables you want to query. * Network access from Genloop. If your database is behind a firewall, allow Genloop's regional IP address. See [Before you begin](/using-genloop/connections/connect-a-database#before-you-begin). ## Connection fields | Field | Description | | ------------------- | -------------------------------------------------------------- | | **Connection name** | A name to identify this connection in Genloop | | **Host** | Hostname or IP address of your PostgreSQL server | | **Port** | The port your PostgreSQL server listens on. Defaults to `5432` | | **Database** | Name of the database to connect | | **Username** | PostgreSQL user Genloop connects as | | **Password** | Password for that user | ## Connect through an SSH tunnel If the database is not reachable directly, reach it through a bastion or jump host. Turn on **Connect via SSH tunnel** in the connection form and complete these fields. | Field | Description | | ----------------------- | ------------------------------------------------------------------------------------------- | | **SSH host** | Public IP or DNS of the bastion. Cannot be `localhost` | | **SSH port** | SSH port on the bastion. Default is `22` | | **SSH username** | User Genloop signs in to the bastion as | | **PEM key** | Your private key file. Sent encrypted with the rest of the credentials | | **Remote bind address** | Database host as seen from the bastion. Use `127.0.0.1` if the database runs on the bastion | | **Remote bind port** | Database port as reached from the bastion | ## Permissions Genloop only reads your data and never writes to your database. You can connect with any user that already has read access to the tables you select. If you want a dedicated read-only user for Genloop, ask your database administrator to set one up. Run this in your PostgreSQL database: ```sql theme={null} GRANT USAGE ON SCHEMA public TO genloop_user; GRANT SELECT ON ALL TABLES IN SCHEMA public TO genloop_user; ``` ## Next steps Back to the full connection flow. View and edit the memory built from your connection. # Redshift Source: https://docs.genloop.ai/using-genloop/connections/databases/redshift Connect an Amazon Redshift cluster to Genloop with cluster endpoint and read-only credentials, optionally through an SSH tunnel, and query your warehouse in natural language. Genloop connects to Amazon Redshift over a direct connection using a read-only user. It reads your schema and the tables you select, and never writes to your cluster. ## Before you connect * A Redshift user with read access to the tables you want to query. * Network access from Genloop. If your cluster is in a VPC, allow Genloop's regional IP address in the cluster's security group. See [Before you begin](/using-genloop/connections/connect-a-database#before-you-begin). ## Connection fields | Field | Description | | ------------------- | ------------------------------------------------------------- | | **Connection name** | A name to identify this connection in Genloop | | **Host** | Your Redshift cluster endpoint | | **Port** | The port your Redshift cluster listens on. Defaults to `5439` | | **Database** | Name of the Redshift database to connect | | **Username** | Redshift user Genloop connects as | | **Password** | Password for that user | ## Connect through an SSH tunnel If the cluster is not reachable directly, reach it through a bastion or jump host. Turn on **Connect via SSH tunnel** in the connection form and complete these fields. | Field | Description | | ----------------------- | ----------------------------------------------------------------------------------------- | | **SSH host** | Public IP or DNS of the bastion. Cannot be `localhost` | | **SSH port** | SSH port on the bastion. Default is `22` | | **SSH username** | User Genloop signs in to the bastion as | | **PEM key** | Your private key file. Sent encrypted with the rest of the credentials | | **Remote bind address** | Cluster host as seen from the bastion. Use `127.0.0.1` if the cluster runs on the bastion | | **Remote bind port** | Cluster port as reached from the bastion | ## Permissions Genloop only reads your data and never writes to your cluster. You can connect with any user that already has read access to the tables you select. If you want a dedicated read-only user for Genloop, ask your database administrator to set one up. Run this in your Redshift database: ```sql theme={null} GRANT USAGE ON SCHEMA public TO genloop_user; GRANT SELECT ON ALL TABLES IN SCHEMA public TO genloop_user; ``` ## Next steps Back to the full connection flow. View and edit the memory built from your connection. # SAP HANA Source: https://docs.genloop.ai/using-genloop/connections/databases/sap-hana Connect an SAP HANA database to Genloop with host, port, schema, and read-only credentials over an encrypted connection. Genloop connects to SAP HANA over an encrypted connection using a read-only user. You provide the schema Genloop should read. ## Before you connect * An SAP HANA user with read access to the schema you want to query. * The **schema** name that holds your tables. * Network access from Genloop. If the server is behind a firewall, allow Genloop's regional IP address. See [Before you begin](/using-genloop/connections/connect-a-database#before-you-begin). ## Connection fields | Field | Description | | ------------------- | --------------------------------------------------------------------- | | **Connection name** | A name to identify this connection in Genloop | | **Host** | Server hostname, for example `xxxxx.hanacloudservices.com` | | **Port** | The port your server listens on. Defaults to `443` for SAP HANA Cloud | | **Schema** | Schema Genloop reads, for example `DBADMIN` | | **Username** | SAP HANA user Genloop connects as | | **Password** | Password for that user | ## Encryption and network The connection is always encrypted and the server's certificate is validated; there is no setting to change this. An SSH tunnel is not used, so make sure Genloop can reach the server directly. ## Permissions Genloop only reads your data and never writes to your database. You can connect with any user that already has read access to the schema you select. If you want a dedicated read-only user for Genloop, ask your database administrator to set one up. Run this in your SAP HANA database: ```sql theme={null} GRANT SELECT ON SCHEMA YOUR_SCHEMA TO GENLOOP_USER; ``` ## Next steps Back to the full connection flow. View and edit the memory built from your connection. # Snowflake Source: https://docs.genloop.ai/using-genloop/connections/databases/snowflake Connect a Snowflake data warehouse to Genloop with your account identifier, warehouse, role, and read-only credentials, and query your data in natural language. Genloop connects to Snowflake over its cloud service using your account credentials. Instead of a host and port, you provide your account identifier, a warehouse for compute, and a role that has read access. ## Before you connect * A Snowflake user and role with read access to the tables you want to query. * A warehouse Genloop can use to run queries. * If your account has network policies, allow Genloop's regional IP address. See [Before you begin](/using-genloop/connections/connect-a-database#before-you-begin). ## Connection fields | Field | Description | | ---------------------- | --------------------------------------------------------------------------------------------- | | **Connection name** | A name to identify this connection in Genloop | | **Account identifier** | Your Snowflake account identifier, from your URL: `account_identifier.snowflakecomputing.com` | | **Username** | Snowflake user Genloop connects as | | **Password** | Password for that user | | **Warehouse** | Warehouse used to run queries | | **Role** | Role Genloop assumes for the connection | | **Database** | Database to query | ## Encryption and network Genloop reaches Snowflake over the internet using an encrypted (HTTPS) connection. There is no SSL setting to configure, and an SSH tunnel is not used. ## Permissions Genloop only reads your data and never writes to your warehouse. You can connect with any role that already has read access to the tables you select. If you want a dedicated read-only role for Genloop, ask your Snowflake administrator to set one up. Run this in Snowflake: ```sql theme={null} GRANT USAGE ON WAREHOUSE your_warehouse TO ROLE genloop_role; GRANT USAGE ON DATABASE your_database TO ROLE genloop_role; GRANT USAGE ON SCHEMA your_database.your_schema TO ROLE genloop_role; GRANT SELECT ON ALL TABLES IN SCHEMA your_database.your_schema TO ROLE genloop_role; ``` ## Next steps Back to the full connection flow. View and edit the memory built from your connection. # Trino Source: https://docs.genloop.ai/using-genloop/connections/databases/trino Connect a Trino coordinator to Genloop with host, catalog, and read-only credentials over HTTP or HTTPS, optionally through an SSH tunnel. Genloop connects to a Trino coordinator over HTTP or HTTPS and queries a single catalog. Trino federates queries across the data sources its catalogs expose. ## Before you connect * A Trino user with read access to the catalog you want to query. * The **catalog** name Genloop should read, for example `hive` or `iceberg`. * Network access from Genloop. If the coordinator is behind a firewall, allow Genloop's regional IP address. See [Before you begin](/using-genloop/connections/connect-a-database#before-you-begin). ## Connection fields | Field | Description | | ------------------- | ---------------------------------------------------------------------------------------- | | **Connection name** | A name to identify this connection in Genloop | | **Host** | Hostname or IP address of your Trino coordinator | | **Port** | The port your Trino coordinator listens on. Defaults to `8443` | | **Catalog** | The catalog Genloop queries, for example `hive` | | **Username** | Trino user Genloop connects as | | **Password** | Password for that user. Leave blank if your cluster does not use password authentication | | **HTTP scheme** | `https` by default. Set to `http` for a non-secure coordinator | ## Encryption (TLS) Leave **HTTP scheme** set to `https` to connect securely. This is the recommended setting. Use `http` only for a coordinator that runs without TLS, for example inside a private network. ## Connect through an SSH tunnel If the coordinator is not reachable directly, reach it through a bastion or jump host. Turn on **Connect via SSH tunnel** in the connection form and complete these fields. | Field | Description | | ----------------------- | ------------------------------------------------------------------------------------------------- | | **SSH host** | Public IP or DNS of the bastion. Cannot be `localhost` | | **SSH port** | SSH port on the bastion. Default is `22` | | **SSH username** | User Genloop signs in to the bastion as | | **PEM key** | Your private key file. Sent encrypted with the rest of the credentials | | **Remote bind address** | Coordinator host as seen from the bastion. Use `127.0.0.1` if the coordinator runs on the bastion | | **Remote bind port** | Coordinator port as reached from the bastion | ## Permissions Genloop only reads your data. You can connect with any user that already has read access to the catalog you connect. Access in Trino is managed through your access-control configuration rather than SQL grants. If you want a dedicated read-only user for Genloop, ask your Trino administrator to set one up. ## Next steps Back to the full connection flow. View and edit the memory built from your connection. # Unstructured Data Source: https://docs.genloop.ai/using-genloop/connections/unstructured-data Upload business documents such as PDFs, Word files, and policies so Genloop learns your KPI definitions, business rules, and processes and applies them when answering questions. Unstructured Data is where you upload business documents: the files that describe how your business works, not what your raw data looks like. Genloop uses them to interpret your data correctly when answering questions. These documents do not create queryable tables. They inform the answers Genloop gives about your connected data. Supported formats: `.pdf`, `.doc`, `.docx`, `.txt`, and other common document types. Maximum 10 MB per file. Unstructured Data tab listing uploaded documents ## What to upload Upload anything that carries business rules, definitions, or context your team uses to interpret data: * Metric and KPI definitions * Policy documents * Playbooks and process guides * Data dictionaries * Reference reports ## Upload a document Open the **Unstructured Data** tab and click **Upload**. Add your file and set who it applies to. You can upload documents at any time. They do not have to be added during onboarding, and you can keep adding them as your business context grows. ## Set who a document applies to Each document has a **Belongs To** scope: * **Organisation** — applies across every workspace in your organisation. * **Team** — applies only within that team's context. ## Manage documents The tab lists each document with its name, description, type, uploader, upload date, and scope. Use the row actions to **edit** its details, **download** the original file, or **delete** it. Search by name to find a document in a long list. ## Next steps See how uploaded documents feed into your workspace context. Add structured data sources alongside your documents. # Upload a data file Source: https://docs.genloop.ai/using-genloop/connections/upload-a-data-file Upload CSV or Excel files to Genloop as queryable data sources, add optional business-context documents, and query them in natural language without a database connection. Uploading a data file turns a spreadsheet into a queryable data source. Genloop maps the file to tables and builds memory on it, the same way it does for a database connection. Use this when your data lives in a file rather than a databaseUploading a data file turns a spreadsheet into a queryable data source. Genloop maps the file to tables and builds memory on it, the same way it does for a database connection. Use this when your data lives in a file rather than a database. Supported formats: CSV and Excel (`.xlsx`, `.xls`). Maximum 10 MB per file.CSV and Excel (`.xlsx`, `.xls`). Maximum 10 MB per file. To begin, open the **Structured Data** tab and click **Upload data file**To begin, open the **Structured Data** tab and click **Upload data file**. Upload data files dialog with an optional business context section Data files and business-context documents are different. Data files become queryable tables. Business-context documents describe how to read that data and do not create tables. You can add both in the same dialog. ## Upload a file Drag files onto the drop zone, or click to browse. Each file becomes a queryable table. You can add more files to the same source later. Expand **Add unstructured business context** and upload documents that explain how your data should be read, such as metric definitions, KPI logic, or business rules. Supported formats: `.pdf`, `.txt`, `.doc`, `.docx`. Maximum 10 MB each. Click **Generate context**. Genloop maps the files to tables, applies any business context, and builds memory on the result. Upload data files dialog with an optional business context section Data files and business-context documents are different. Data files become queryable tables. Business-context documents describe how to read that data and do not create tables. You can add both in the same dialog. ## Upload a file Drag files onto the drop zone, or click to browse. Each file becomes a queryable table. You can add more files to the same source later. Expand **Add unstructured business context** and upload documents that explain how your data should be read, such as metric definitions, KPI logic, or business rules. Supported formats: `.pdf`, `.txt`, `.doc`, `.docx`. Maximum 10 MB each. Click **Generate context**. Genloop maps the files to tables, applies any business context, and builds memory on the result. ## After the upload The file appears in your **Structured Data** list with a processing status. Once processing completes, you can query it in Chat like any connected database. ## Next steps Manage business documents that inform Genloop's answers. Connect a live database for larger or changing datasets. # Context Hub Source: https://docs.genloop.ai/using-genloop/context-hub Review and edit everything Genloop knows about your business — tables, KPI definitions, playbooks, and team preferences — from one central Context Hub. Context Hub is the knowledge layer that makes Genloop accurate for your specific business. It holds everything the platform has learned about your data: your table structures, your metric definitions, your processes, and how your team prefers to work. The more complete your Context Hub, the more precise your answers get. You can view, edit, and add to it at any time. Table Data ## What is in Context Hub Context Hub is organized into four layers: **Table Data** Every connected table and column, with semantic descriptions of what each one means, how tables relate to each other, and how Genloop should use them when answering questions. This is generated automatically when you connect your data and can be edited by workspace owners and admins. **Playbooks** Your business processes, workflows, and operational rules. Adding playbooks helps Genloop understand not just what the data says, but how your business actually operates. **KPIs** Your metric definitions — how you calculate revenue, churn, conversion rate, or any other number that matters to your team. When Genloop knows how you define a KPI, it applies that definition consistently across every answer. **User Preferences** How your team prefers to work with data — default filters, preferred ways to slice numbers, recurring context that applies across most questions. ## Editing context Click the edit icon on any item to update its description or definition. Workspace owners and super admins can edit all context layers. Regular members can view but not edit. Context Hub was previously called Memory. If you are coming from an earlier version of Genloop, everything that was in Memory is now here. ## Next steps Ask a question, then use Remember This to add what Genloop got right. Approve memory updates and validation requests from your team. # Data Apps Source: https://docs.genloop.ai/using-genloop/data-apps Turn a Liveboard or a Chat question into a Data App: a report or presentation built on live, governed data, with interactive charts and numbers that always match your warehouse. A Data App turns the analysis already on a Liveboard, or a question you ask in Chat, into a report or a deck. No one has to spend an evening packaging screenshots into slides. ## Creating a Data App Build one from a Liveboard, or ask for one directly in **Chat**: describe what you want ("Build me a QBR deck for this quarter") and Genloop compiles it from your live data. A generated Data App includes an executive summary, key metric and chart insights, the charts from your tiles, and AI-generated highlights on anomalies and what needs attention first. It renders as a full **Report** or **Presentation**, can be downloaded as PDF, and is listed under the board's **Reports** view. ## Built on live, governed data The charts inside are interactive, not static screenshots. Whoever you share it with can explore the numbers, not just read them. And because a Data App runs on the same governed layer as the rest of Genloop, the numbers in it match the numbers in your warehouse. No copy-paste drift between the dashboard and the boardroom. ## Scheduling a Data App Any Data App you schedule for a recurring cadence shows up in [Automations](/using-genloop/automations), where you can track its next run, edit it, or remove it. Your weekly business review, your QBR, your customer success deck: compiled and delivered without you opening a slide editor. ## Next steps Build the Liveboard tiles a Data App pulls from. Put a recurring Data App on a schedule, or on a condition. # Liveboards Source: https://docs.genloop.ai/using-genloop/liveboards Turn saved Chat answers into live dashboards. Build Liveboards with KPI tiles that refresh with your data, filter and share them, ask questions about them, and turn them into Data Apps. A Liveboard is a live dashboard built from answers you save in Chat. Each tile is a pinned question (a KPI number, a chart, or a table) that stays connected to your data and re-runs when you refresh, so the board always reflects the latest numbers. You can keep multiple Liveboards, each organized around a topic, team, or business area. On top of the tiles, Genloop layers AI highlights, board-wide filters, a scoped chat, and scheduled reports. A Liveboard is somewhere you *work with* your data, not just look at it. Personal Liveboards in Genloop ## Finding your Liveboards Go to **Liveboard** in the sidebar. Your boards are organized across four tabs: * **Starred**: boards you have favorited for quick access * **Personal**: boards you created * **Team**: boards published to your workspace * **Shared**: boards other people have shared with you Use **Type to search...** to filter boards by name, and click the **star** on any card to add it to your Starred tab. Each card shows a preview chart and the **Last Edited** date; shared boards also show the owner. **Create New Liveboard** appears on the Personal tab. Team and Shared Liveboards are a paid-plan feature. On the Free plan you can view them but cannot publish or create them until you upgrade. ## Creating and sharing a Liveboard Click **Create New Liveboard** to start a blank board with a name and description, or [save an answer from Chat](#adding-tiles-from-chat) to spin up a new board on the fly. Once you have a board, you can share it two ways: * **Share with people**: Grant specific teammates access at one of three permission levels: * **View**: read the board and its data * **Edit**: rename tiles and adjust filters * **Full**: edit tile SQL, remove tiles, and manage sharing * **Publish to Team**: Make the board available to your whole workspace under the Team tab. Each viewer sees the board through their own data access. The same tile can show different numbers to different people depending on what they are permitted to see. Team admins and data experts get extra sub-tabs on the Team tab for **Share Requests** and **Validation Requests**. ## Adding tiles from Chat When an answer comes back in Chat, click **Save to Liveboard**. A dialog lets you search for an existing board, create a new one inline, or select several boards at once, then **Add to Liveboard**. Toggle between chart and table view before you save to control how the tile first appears. Save a Chat answer to a Liveboard The saved tile stays connected to its underlying query, so it re-runs against current data when the board refreshes. ## Working with tiles A Liveboard holds two kinds of tiles: * **Metrics**: a single KPI value, shown in a grid at the top of the board * **Graphs**: charts or tables, shown below, each with a **Table / Visualisation** toggle Genloop picks the chart type that fits the question: bar, line, funnel, heat map, stacked column, area, and more. On any tile, open the visualisation and switch chart types, adjust the axes, or change the breakdown to see the same data differently. Switch Chart Hover a tile to reveal a drag handle and **reorder** it; the new arrangement is saved for the board. Open a tile's **...** menu for the full set of actions: * **Edit name**: rename the tile * **View / Edit SQL**: inspect or change the query behind the tile (see below) * **Download Image**: export the chart as a PNG * **Export CSV**: download the full data, or the table- or chart-filtered rows * **Copy Graph ID / Table ID**: copy an identifier (optionally with the current filters) to embed the tile elsewhere * **Remove**: delete the tile from the board What you can do depends on your permission level: renaming needs **Edit**, while editing SQL and removing tiles need **Full**. ### Viewing and editing a tile's SQL From a tile's menu, choose **View / Edit SQL** to see the exact query behind it. If you have permission, you can adjust the SQL and re-run it. Genloop also exposes the tile's filterable columns as editable **toggles**, while structural parts of the query stay fixed. So you can change *what* the tile shows without breaking *how* it is built. ### Tiles with multiple charts Some questions are too big for one chart. Ask something like "how did revenue move across channels and categories this quarter" in Chat, and Genloop weaves more than one chart into a single answer instead of making you ask three follow-ups. Save that answer to a Liveboard, and the tile keeps every chart together, exactly as it appeared in Chat. ## Highlights and suggested questions At the top of each board, Genloop surfaces a **Highlight**: a short, AI-generated summary of what stands out across the tiles (for example, a metric that moved sharply), along with a few **suggested questions** you can click straight into a scoped chat. Genloop also flags notable **anomalies** in the underlying data, which you can ask about directly. Highlights are personalized to the data *you* can see and regenerate when you refresh the board or change its filters. A Liveboard with its Highlight summary, suggested questions, metric tiles, and a chart ## Filtering a Liveboard Two board-level filters let you slice every tile at once: * **Date filters**: Pick a window (Today, Last 7 / 30 / 90 days, or a custom range) applied across the board's date columns. * **Filters**: Set values for the variables in your tiles (for example, *Loan Status* or *Region*) from the **Add Filters** sheet. Both use **Apply Filters** and **Reset**, and show a badge with the number of active filters. Filters are saved per person, so your view of a shared board does not change what anyone else sees. Applying a filter refreshes all tiles and updates the highlights. The Add Filters panel with per-variable value selectors ## Keeping data live Tiles are not static snapshots. Click **Refresh** to re-run every tile against current data and regenerate the highlights. There is a short cooldown between refreshes, and the button shows when the board was last refreshed. ## Asking questions about a Liveboard A Liveboard is not just something to read. You can ask questions directly on top of it. **Ask Liveboard** opens a conversation that already knows what you are looking at, so you can dig into a number without re-explaining your data or starting from a blank Chat. There are three ways to start: * Click **Ask Liveboard** in the board header to ask about the **whole dashboard**. * Use the **Ask** action on any tile to focus on that **single metric or chart**. * Click one of the **suggested questions** in a Highlight to jump straight in with that question. The conversation opens in a **panel docked beside your board**, so your tiles stay in view while you investigate. On a fresh thread you will see **Ask Liveboard** with a prompt to *"Investigate why something changed"* and a few **Quick Asks** to get you going. A context chip shows what your question is scoped to (the board name, or *Item: \* for a single tile), and you can remove it to widen the scope. The Ask Liveboard panel docked beside a board, with Quick Asks and a context chip ### What makes it different from Chat Ask Liveboard runs on the same engine as [Chat](/using-genloop/chat), so you get the same verified answers, but grounded in the dashboard in front of you: * **It answers on exactly what you see.** Questions run against the current data behind your tiles, including any filters you have applied and the way each chart is currently set up. You do not pick a database or restate the context. * **Same depth as Chat.** Every answer comes with its **query plan** and charts, you can **view or edit the SQL**, and Genloop will ask **clarifying questions** or suggest **follow-ups** as the conversation builds. * **A running thread per board.** Each Liveboard keeps its own Ask history, so you can reopen an earlier investigation or start a new one from the panel, and the answer keeps generating even if you close the panel to look around, ready when you come back. When you want more room or the full toolset, click **Open in Chat** to continue the same conversation in the main Chat view. ## Data Apps: reports and presentations Turn any Liveboard into a document or a deck built on live data. See [Data Apps](/using-genloop/data-apps) for how to create, format, and schedule one. ## Next steps Turn this Liveboard into a report or presentation. Put a recurring Data App on a schedule, or on a condition. # Review Center Source: https://docs.genloop.ai/using-genloop/review-center Govern what Genloop learns from your data team in the Review Center: approve validation requests, review memory updates, and triage user feedback in one place. Review Center is where continuous learning is governed. When users flag answers in Chat, request validation, or when Genloop proposes new context to learn, it all lands here for your data team to review. It is the control layer between what Genloop sees in your data and what it actually learns — making sure your team stays in the loop on every update to the platform's intelligence. Review Center ## What lands in Review Center Review Center has three tabs: **Validation Requests** When a user clicks **Request Validation** on a Chat answer, it creates a ticket here. Your data team can see the original question, the SQL Genloop generated, and the answer it returned. They can review the logic, mark it as correct or incorrect, and approve it. That approval feeds directly back into the platform as a learning signal. Validation requests are split into Chat Requests and Liveboard Requests. **User Ratings** Thumbs up and thumbs down feedback from Chat is aggregated here. Your data team can use this to spot patterns — which questions are consistently getting wrong answers, which areas need more context. **Memory Updates** When Genloop learns something new from interactions and proposes a context update, it surfaces here for review. Your team can approve or reject the update before it becomes part of Context Hub. ## Who uses Review Center Review Center is primarily for data team members and workspace admins. It is the main surface for governing what the platform learns and catching issues before they become repeated errors. The average days to resolution metric at the top of the page tracks how quickly your team is turning around validation requests. A high backlog here means users are waiting on answers they are not confident in. ## Next steps See where approved memory updates end up. Request Validation and give feedback directly from an answer.