> ## Documentation Index
> Fetch the complete documentation index at: https://docs.wisdom.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# FAQ

Answers to common questions from WisdomAI users, grouped by topic.

## AI Model

<AccordionGroup>
  <Accordion title="What AI model does WisdomAI use?">
    WisdomAI isn't tied to a single model. On SaaS Hosted, WisdomAI uses a combination of models from OpenAI and Anthropic under Zero Data Retention (ZDR) agreements, so prompts and completions aren't retained by those providers. On Cloud-Prem (VPC) and Customer On-Prem, you can bring your own LLM (BYO-LLM) and point WisdomAI at your own on-prem or private-cloud model endpoint instead. See [Deployment Models](/deployment/deployment-models).
  </Accordion>

  <Accordion title="Does WisdomAI have its own AI model?">
    No. WisdomAI doesn't train or ship a foundation model of its own. On SaaS Hosted, it calls out to OpenAI and Anthropic models under Zero Data Retention agreements.
  </Accordion>

  <Accordion title="Can I bring my own AI model to use with WisdomAI?">
    Yes, on Cloud-Prem and Customer On-Prem. This is Bring Your Own LLM (BYO-LLM): point WisdomAI at your own on-prem or private-cloud model endpoint, with no hard dependency on any single provider. Not available on SaaS Hosted, where WisdomAI operates model access for you. See [Deployment Models](/deployment/deployment-models).
  </Accordion>
</AccordionGroup>

## Data Sources

<AccordionGroup>
  <Accordion title="What database does WisdomAI run on?">
    WisdomAI isn't a database itself. It's an analytics layer that connects to and queries your own databases and warehouses (Snowflake, BigQuery, Redshift, Databricks, PostgreSQL, SQL Server, and more). For its own platform metadata, audit logs, and, on self-hosted deployments, the vector/lexical index, WisdomAI runs on PostgreSQL and OpenSearch. See [Supported Data Sources](/getting-started/connect-data-sources/supported-data-sources).
  </Accordion>

  <Accordion title="Can WisdomAI write data back to a source like S3?">
    Regular data-source connections, including Amazon S3, are read-only. To write data out, use an Export node inside an Agent: it sends a dataframe the agent produced to a connected S3 or Azure Blob Storage destination as a CSV file, provided that connection has write access. See [Nodes Reference: Export nodes](/setting-up-wisdom-ai/agents/nodes-reference#export-nodes) and [Connect Unstructured Repositories or Datasets](/improve-wisdom-ai-responses/connect-unstructed-repositories-datasets).
  </Accordion>

  <Accordion title="Do I need a data warehouse to use WisdomAI?">
    Not a requirement. WisdomAI connects directly to a wide range of databases, warehouses, and SaaS sources, so a dedicated warehouse isn't necessary just to use it. That said, consolidating data into a warehouse (Snowflake, BigQuery) is a common pattern at scale, since it simplifies semantic modeling and access control across many source systems. It's a data-architecture choice independent of WisdomAI, not a platform requirement.
  </Accordion>
</AccordionGroup>

## Cross-Domain Queries

<AccordionGroup>
  <Accordion title="Can I query or join data across multiple domains?">
    Yes. This is what Domain Groups and Auto Domain Selection are for. Group two or more domains so a conversation can draw on tables across all of them, or turn on Auto mode to have WisdomAI route each question to the right domain or group automatically. See [Set Up and Use Domain Groups & Auto Domain Selection](/setting-up-wisdom-ai/domain-groups-how-to).
  </Accordion>
</AccordionGroup>

## Embedding

<AccordionGroup>
  <Accordion title="Is there a size limit for the JWT I pass in the embed URL?">
    There's no WisdomAI-imposed cap on the token itself. Two things matter more in practice: embedded session JWTs already expire after 1 hour by design, and a token with many custom claims can hit URL-length limits enforced by browsers or intermediate proxies, typically in the low thousands of characters. Keep claims minimal and rely on the built-in `postMessage` refresh flow instead of a long-lived, claim-heavy token. See [Session Management](/integrations/embeddings/session-management).
  </Accordion>
</AccordionGroup>

## Usage Limits

<AccordionGroup>
  <Accordion title="What are the usage limits for chat and conversation analytics?">
    A few concrete ones exist today: chat file attachments are capped at 20MB per file (PNG/JPEG/WebP, PDF, and common text formats); result tables loaded into the browser support Instant Filters on up to 10,000 rows; and large SQL results are automatically truncated before being summarized so the model reasons over complete rows.
  </Accordion>

  <Accordion title="How many domains can I have?">
    There's no hard platform-enforced cap. The one built-in behavior change: past 10 domains, Auto Domain Selection requires configuring a Domain Greenlist (the subset WisdomAI is allowed to auto-route to). Any plan-level quota is set by your subscription. Check with your account team for your specific limit.
  </Accordion>
</AccordionGroup>

## Billing

<AccordionGroup>
  <Accordion title="Does my subscription include infrastructure costs?">
    Depends on the deployment model. SaaS Hosted: yes. WisdomAI manages all compute, storage, and networking, with no separate infrastructure cost to plan for. Cloud-Prem / Customer On-Prem: the subscription covers the platform itself, while the underlying cloud spend (AWS/GCP/Azure) or on-prem hardware is provisioned and billed separately by you, or offset against existing cloud commitments. See [Deployment Models](/deployment/deployment-models).
  </Accordion>
</AccordionGroup>

## Integrations

<AccordionGroup>
  <Accordion title="Can I connect WisdomAI to Microsoft Teams?">
    Not today. The chat-app integration currently supports Slack only. Microsoft Teams isn't supported. If it's a priority for your organization, flag it to your account team for the roadmap. See [Connect to Slack](/integrations/connect-to-slack).
  </Accordion>
</AccordionGroup>

## MCP

<AccordionGroup>
  <Accordion title="Can I authenticate to the WisdomAI MCP server with my identity provider instead of OAuth?">
    If your admin has enabled it, you can authenticate to the WisdomAI MCP Server with an access token from your organization's identity provider (e.g. Azure AD / Entra ID) instead of standard OAuth sign-in or an API key. Ask your WisdomAI admin to enable IdP-based auth and share your MCP URL. See [WisdomAI MCP Server: Authenticate](/integrations/mcp-server#step-3-authenticate).
  </Accordion>

  <Accordion title="How do I connect Glean or another internal tool to the WisdomAI MCP server?">
    Any MCP-compatible client connects the same way: URL, Streamable HTTP transport, and OAuth (or an API key). There's no Glean-specific guide because none is needed. Once Glean, or any internal tool, supports connecting to a remote MCP server, it uses that same generic configuration. See [WisdomAI MCP Server: Other clients](/integrations/mcp-server).
  </Accordion>

  <Accordion title="What are the limitations of using WisdomAI with MCP?">
    WisdomAI has two distinct MCP capabilities, each with its own limits: (1) the hosted WisdomAI MCP Server, which lets clients like Claude, ChatGPT, Cursor, and VS Code query your domains. It doesn't yet support creating live apps, building agents, or uploading files (use the app for those); and (2) connecting an external MCP server as a data source inside WisdomAI. This scans metadata and fetches data, but isn't a general two-way integration. See [WisdomAI MCP Server: Not available via MCP](/integrations/mcp-server#not-available-via-mcp).
  </Accordion>
</AccordionGroup>

## Single Sign-On (SSO)

<AccordionGroup>
  <Accordion title="Can I set up SSO with Okta?">
    Yes. Okta is one of the identity providers supported under Bring Your Own IdP (BYO IdP) for enterprise SSO, alongside Entra ID, on Cloud-Prem and Customer On-Prem. Once SSO is live, use SSO Groups to map Okta groups or JWT claims to WisdomAI roles. See [Role Mapping & SSO Groups](/manage-account/role-mapping-and-sso-groups).
  </Accordion>

  <Accordion title="How do I set up OAuth-based SSO?">
    This is the same feature as Okta SSO above. OAuth/OIDC-based SSO is covered by the same BYO IdP setup (Okta or Entra ID). See [Role Mapping & SSO Groups](/manage-account/role-mapping-and-sso-groups).
  </Accordion>
</AccordionGroup>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.