AgentQ FAQ

Common questions about AgentQ, the AI assistant built into the Qualytics platform: what it is, how to set it up, conversations, responses, limits, MCP clients, and security. For something that is not working rather than not clear, see Troubleshooting.

General

What is AgentQ?

AgentQ is the AI-powered assistant integrated into the Qualytics platform. It uses the Model Context Protocol (MCP) and your configured AI provider to give you natural-language access to shared, governed context and the actions available through your Qualytics permissions. You can ask it to explore datastores, create quality checks, investigate anomalies, run operations, send notifications, create tickets, and perform other supported tasks. Review its responses and resulting changes before relying on them.

How does AgentQ work?

AgentQ connects to your configured AI provider and uses a set of MCP tools to interact with the Qualytics platform on your behalf. When you send a message:

1. A scope guardrail checks whether the request relates to data quality or the Qualytics platform.
2. Your configured LLM interprets the request using your prompt and relevant Qualytics context.
3. AgentQ selects from the Qualytics actions available to your account.
4. It performs those actions one at a time or in sequence as needed.
5. Results are streamed back with progress indicators for each step.

Review the completed steps, generated explanations, and any platform changes before using the result.

Is my data sent to the LLM?

AgentQ sends your prompt, relevant asset context, and results from the tools used to fulfill your request to your configured AI provider. What a tool can return is limited by its capabilities and your Qualytics permissions. Those results are metadata and summaries: asset names and schemas, profile statistics, quality check definitions, anomaly counts and messages, and operation status. AgentQ's tools do not return rows from your source systems, and point you to the Qualytics interface for record-level inspection instead. Field profile statistics can include values derived from your data, such as histogram buckets and minimum and maximum values, and masked fields are excluded from those. The model does not connect directly to your source systems, and no tool returns data beyond what the requesting user is authorized to access in Qualytics. Your provider handles the information it receives according to its own privacy and data processing policies.

Will AgentQ act on requests outside of data quality?

AgentQ is designed to decline requests outside data quality and the Qualytics platform. A scope guardrail checks whether a request relates to data quality, governance, databases, anomalies, transformations, or Qualytics. Because the guardrail uses an LLM, it can occasionally misclassify a request. Rephrase a relevant request with more specific data quality context if needed.

Setup

What AI providers are supported?

AgentQ supports the Qualytics-managed provider (Beta) and a range of external providers, including OpenAI, Anthropic, Google Gemini, Amazon Bedrock, Groq, Mistral, DeepSeek, Cohere, Ollama, OpenRouter, and Perplexity. See the external-provider list with example models on the Supported AI Providers page.

Do I need my own LLM API key?

No. You can choose the Qualytics provider (Beta), which does not require an API key, model choice, or custom endpoint. You can also connect a supported external provider with your own credentials.

How is the provider configuration validated when I save it?

Qualytics sends a minimal request to the selected provider to confirm connectivity before saving. For an external provider, this test also validates the supplied credentials. The Qualytics-managed provider is tested with the deployment's managed connection and does not require your organization to supply an API key.

What is the Base URL field for?

The Base URL is an optional custom endpoint for OpenAI-compatible providers that self-host or proxy models. Use it for providers like Ollama (local), OpenRouter, LiteLLM proxy, or any other provider that exposes an OpenAI-compatible API at a custom URL.

Can AgentQ search the Qualytics User Guide?

AgentQ does not provide general web search. It searches a copy of this User Guide that ships with your Qualytics release, whichever AI provider is configured, so answers always match the version you are running. It returns up to 4 of the most relevant sections per search, and the answer quotes or paraphrases from them, with a link that opens the guide hosted by your own deployment. Review the linked documentation before relying on an answer for a business-critical decision. See Documentation Answers for the full behavior.

Can multiple users configure different AI providers?

No. The AgentQ configuration is deployment-wide. A single Manager or Admin selects the provider for all users with the Member role or higher. External providers also require a model and credentials. The Qualytics-managed provider does not. Only one active provider configuration exists at a time per deployment.

What happens if a Manager changes the AI provider?

Existing conversation history is preserved. Sessions are stored independently of the AgentQ configuration. The new provider takes effect immediately for every user on their next message. Different models may behave differently for the same prompts, so results may vary across providers.

What happens if I don't configure an AI provider?

Without an AI provider connected, the AgentQ page shows a preview of what it does, headed Ask about your data. Act on the answer. A Manager or Admin sees Easy setup and Custom setup cards that lead straight into the setup. Everyone else sees the note An administrator connects a provider once, and it turns on for everyone here. The floating chat stays hidden, and the assistant panel in the check and template editors shows the same card. The platform is fully functional otherwise. Follow the Add Integration guide to get started.

Opening AgentQ and Conversations

How do I open AgentQ?

Use any of these entry points:

• Click AgentQ in the left sidebar for the full-page chat interface.
• Click the AgentQ mascot, which sits in the bottom-right corner of any page until you drag it elsewhere. Press the Q key to toggle the chat from anywhere on the page (except while typing in an input). If the mascot is tucked away at the bottom edge, click its peeking head to bring it back.
• Open the AgentQ assistant inside a quality check or check template dialog to draft settings from a plain-language requirement. Apply the proposal to the form, review the highlighted fields, and save only when the result is correct.

What are the smart suggestions?

When you start a new conversation, AgentQ generates 3 personalized prompt suggestions using the LLM, based on your actual containers and active anomalies. These highlight common workflows like investigating anomalies, creating quality checks, and analyzing trends. Click any suggestion to use it as your opening message.

Suggestions are hidden when you open AgentQ from a page where context is already injected (e.g., from an anomaly page).

How does context injection work?

When you open AgentQ from a page that has relevant data (a datastore, container, field, quality check, or anomaly), AgentQ automatically receives that asset's identity as context. This context appears as a badge above the input box (icon + asset name). You can then ask questions like "Explain this anomaly" or "What checks exist here?" without specifying which asset you mean.

The context is included with your first message and is visible through the Context action button after the message is sent.

Can I have multiple conversations?

Yes. AgentQ supports full session management. You can start new conversations at any time, switch between chats from the history sidebar or the floating chat dropdown, search by title and by message content, and rename, archive, restore, or delete chats.

Does AgentQ remember previous conversations?

Yes. Each session maintains its message history. In long sessions, AgentQ keeps recent messages in detail and summarizes older context so the conversation can continue within the model's limits.

If you notice AgentQ losing context in a very long conversation, starting a new session for a fresh context window is the most reliable approach.

What happens if I navigate away while AgentQ is generating?

The response continues streaming in the background. When you return to that session, the completed (or in-progress) response is waiting for you. You can see which sessions are still generating from the loading indicators in the session list.

Can I share a conversation with another user?

Not directly. Conversations are private to your user account. To share findings, use the Export as PDF button below any assistant message to download a formatted PDF you can share externally.

How long are conversations stored?

Conversations persist indefinitely until you delete them. Archived conversations are also kept until explicitly deleted. See Delete a Conversation for instructions on permanent removal.

What languages does AgentQ support?

AgentQ can understand and respond in any language your configured LLM model supports. The platform interface is in English, but you can write prompts in your preferred language and AgentQ will respond in kind.

Actions and Files

Can AgentQ make changes to my platform configuration?

Yes, and the list is broad. AgentQ can create, update, and delete quality checks and check templates; create, update, and delete computed assets (tables, files, joins, fields); edit asset metadata such as names, descriptions, and tags; update and archive anomalies; add and edit comments; promote assets across datastores; trigger operations (sync, profile, scan, export, materialize) and manage running ones; create and manage schedules; send notifications; and create tickets. Every one of these stays within your Qualytics permissions, and some require the Manager or Admin role. Changes take effect as soon as AgentQ makes them and there is no undo, so confirm the target and the impact before asking for anything destructive. Changes are recorded on the changed item's own Timeline, and a save made after applying an AgentQ proposal is marked with AgentQ co-authorship alongside your user identity.

Can I export AgentQ responses?

Yes. Click the Export as PDF button below any assistant message to download a formatted PDF. You can also copy any response to your clipboard.

Can I upload files to AgentQ?

Yes, when your active AI provider supports file attachments. The chat input shows an Attach icon for PDF, Word (.doc, .docx), Excel (.xls, .xlsx), CSV, TSV, JSON, XML, plain text, and Markdown files. Limits: one file per message, up to 20 MB per file. The button is shown for Qualytics, Anthropic, Google Gemini, Google Vertex AI, Amazon Bedrock (Claude models), OpenAI, Azure OpenAI, and Heroku. The Qualytics provider reads a narrower set: PDF, Word .docx, Excel .xlsx, and text-based files, but not the legacy .doc and .xls formats, PowerPoint, or images. See Attach a File for full details.

Can I talk to AgentQ instead of typing?

Yes, when your active AI provider can transcribe speech: OpenAI, Groq, Google Gemini, Google Vertex AI, and the Qualytics provider. Click Talk to AgentQ in the chat input and speak; your words are typed into the input as you go, and nothing is sent until you click Send message. See Voice Input.

Why don't I see the microphone?

The microphone is hidden when the active AI provider cannot transcribe speech, which is the case for Anthropic, Amazon Bedrock, Azure OpenAI, and the other OpenAI-compatible providers. It is also hidden for OpenAI set to the Chat Completions API style and for Groq with a custom Base URL, and in the assistant panel of check and check template dialogs. See Supported AI Providers.

Is my voice recording stored?

The audio is not. The recording goes only to the AI provider connected for your deployment, which turns it into text, and the audio is never stored or logged. The text is saved only once you send it, in the chat history like any typed message. A single dictation stops after two minutes.

Responses and Reliability

What happens if AgentQ makes a mistake?

Ask AgentQ to propose or apply a specific correction in the same conversation. For example: "That check is wrong. The field should be not-null, not unique. Please update it." Whether AgentQ can make the correction depends on the available tools and your permissions. Review the completed tool steps and the updated object to confirm the result.

AgentQ does not provide a universal undo. If a change was already applied, use the platform interface or ask AgentQ to perform a specific follow-up action that its tools support, such as deleting a newly created check. Confirm the target and impact before applying a destructive action, then verify the resulting platform state.

What does the "Answer basis" note under a response mean?

It states how the answer was produced. Direct tool result means a tool directly returned the requested fact or performed the requested action. Model-synthesized answer means the model wrote the answer rather than a tool returning it: AgentQ combined, compared, calculated, or inferred from one or more tool results, reasoned from the evidence a trust assessment or guided workflow supplied, or answered without running a tool at all. It also appears on the routine case of a reply that used more than one listing tool or paged one listing more than once, so it is a common label rather than a warning. Could not verify, no direct tool means the available tools cannot support a reliable or complete answer. A Limitations line is added when something keeps the answer from being complete, such as a tool reporting more results than it returned. See Answer Basis for details, including the replies that carry no note.

Can AgentQ tell me whether a table is trustworthy?

Yes. Ask whether an asset can be relied on, ideally saying what you intend to use it for, and AgentQ assesses it on the spot with no setup. The verdict is anchored to the asset's Quality Score and explains which dimensions drive it, which anomalies are active, how fresh the data is, and whether upstream assets are healthy. An asset that has never been scored gets an explicit insufficient evidence answer instead of a low rating. See Trust Assessment.

Why does AgentQ ask before running some analyses?

When no tool answers a request directly and deriving an answer would mean fetching and comparing large amounts of data, AgentQ explains the gap and asks for confirmation before running the slower best-effort analysis. The analysis does not run until you agree, so an expensive and potentially incomplete comparison is not started on your behalf. Reply yes, or another short confirmation such as proceed or go ahead, as your next message in that conversation. See Confirmation Before Best-Effort Analyses for the wordings it accepts and when the confirmation is mandatory.

Limits and Cost

Are there usage limits?

Yes. To ensure fair usage and predictable costs:

If you hit the rate limit, wait a moment and try again. If you consistently exceed token limits on a single request, try breaking it into smaller focused steps.

See Limits for the full set of constraints, including timeouts, listing caps, and SQL restrictions.

How much does it cost to use AgentQ?

Costs depend on the configured provider and your organization's commercial terms. With an external provider, your organization pays that provider based on usage. The Qualytics-managed provider does not require a separate LLM account or API key. Ask your deployment administrator or Qualytics contact which terms apply to your environment.

To reduce costs, use specific, focused prompts and scope requests to a particular datastore or container rather than platform-wide queries. See Best Practices for detailed guidance.

MCP Integration

What is MCP?

The Model Context Protocol is an open standard that enables AI assistants to securely connect to external data sources and tools. Qualytics implements an MCP server that exposes its data quality functionality to any compatible client.

Can I connect external AI clients to Qualytics?

Yes. Beyond the built-in AgentQ, you can connect ChatGPT, Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, Amazon Q Developer, and any MCP-compatible client directly to the Qualytics MCP server using your Personal API Token. See the Connecting External AI Clients guide for step-by-step instructions for each client.

Which external AI clients are supported?

Qualytics provides step-by-step setup guides for the following MCP clients:

Any other MCP-compatible client that supports Streamable HTTP transport can also connect. See the Connecting External AI Clients guide for full instructions.

Do external clients have the same capabilities as AgentQ?

External MCP clients share the same tool set as AgentQ. They connect to the same MCP server and have access to the same tools for exploration, quality checks, transformations, anomalies, operations, and integrations. However, features specific to the Qualytics UI (context injection, smart suggestions, session management, PDF export) are only available in the built-in AgentQ chat.

What is the difference between AgentQ, the MCP server, and the agentic endpoints?

All three use the same underlying MCP tools and capabilities. They differ only in how you access them.

What tools are available?

AgentQ and external MCP clients use the same tool catalog. The data and actions available in a session depend on your Qualytics permissions and the relevant feature configuration. The table below highlights the most common tools. See the Tool Catalog for the full list and the category of data each tool shares.

Security

What safeguards apply when using AgentQ in production?

AgentQ includes safeguards that reduce risk, but they do not replace human review or your organization's governance requirements. Suitability depends on your use case, configured AI provider, permission model, and data-handling requirements. Use least-privilege access and verify proposed and completed changes, especially for business-critical or regulated data.

• Scope guardrail: Requests outside data quality and Qualytics are declined.
• Query restrictions: Computed asset queries are validated. Only SELECT statements and CTEs are permitted. Statements that insert, update, delete, or change database objects are blocked.
• Context protections: Content returned by Qualytics tools is processed before it is sent to the LLM.
• Usage limits: Request, concurrency, timeout, and token limits help contain resource usage.
• Co-authorship tracking: changes AgentQ makes on your behalf are recorded on the changed item's Timeline with AgentQ named alongside you.
• Permission enforcement: AgentQ respects the same roles and team permissions as the rest of the platform. It can only access data and actions available to your user account.

AgentQ sends prompts, relevant context, and tool results to your configured AI provider. Review that provider's privacy, retention, and data-processing terms before enabling AgentQ for sensitive data. On the Qualytics-managed provider the data goes to the Qualytics gateway instead, which processes it on Amazon Web Services (AWS) as described in the notice shown when you select that provider.