AI Assistants Overview¶
Using AI assistants
AI assistants are a core capability available with every module combination. The actions they take follow the same module requirements as the rest of the platform: when a request needs a module that is not enabled, the assistant explains the requirement instead of acting.
AI assistants let you explore your data quality, investigate anomalies, create quality checks, and run operations by asking in plain language. You can chat with AgentQ, the assistant built into Qualytics, or connect an AI client you already use, such as ChatGPT, Claude, or Cursor, through the MCP server. This section covers:
- What AgentQ is, how it works, and how to connect the AI provider that runs it
- How to use and manage AgentQ chats, access controls, and audit step by step
- What the MCP server is, how to connect each external AI client, and what a client can do on your behalf
Permissions
Every user can chat with AgentQ once an AI provider is connected. Connecting or changing the provider and setting the access level requires the Manager or Admin role, and reading the AgentQ audit requires Admin. Every user can connect an external AI client through MCP with their own Personal API Token. Whatever an assistant does on your behalf follows your role and team permissions. See AgentQ Permissions and MCP Permissions for the full matrices.
AgentQ¶
Start here for the assistant built into Qualytics. This section covers how AgentQ works, the AI providers that can run it, access controls, and audit, along with the how-tos for connecting it and managing chats, the AgentQ API, and FAQ.
-
Getting Started with AgentQ
What AgentQ does, where you can use it, and every page in its section.
-
Introduction
What this section covers, and the order to read it in.
-
How It Works
Where AgentQ appears, how a turn runs, what it is allowed to see, and its built-in features, including description suggestions.
-
The Chat Interface
Every control in the full-page and floating chat, and what the input accepts.
-
Conversations, Responses & Context
How to write prompts, read AgentQ responses, and work with context-aware chats.
-
AgentQ Limits
Rate limits, token usage, timeouts, SQL constraints, and scope constraints.
-
AgentQ in Action
Worked examples of AgentQ, its tools, and the steps shown in chat.
-
Tool Catalog
Every tool AgentQ can call, what each one does, and what it shares with the model.
-
Semantic Reporting
The structured query interface AgentQ uses to count, filter, and inspect platform resources.
-
Supported AI Providers
Every provider you can connect and which ones support file attachments.
-
Amazon Bedrock Authentication
The three ways to authenticate to Bedrock, and what an IAM role setup expects.
-
Access Controls
How an administrator chooses what chat may share with the model, and what each level unlocks.
-
AgentQ Audit
What each audit entry holds, how cost estimates work, and what the period summary reports.
-
Examples
Worked scenarios showing what you ask AgentQ, what it does, and the shape of the answer.
-
Best Practices
Prompt design, cost management, guardrail behavior, rate limits, and async operation patterns.
-
Permissions
The user roles behind chatting with AgentQ, configuring it, and reading the audit.
-
Add Integration
Connect AgentQ from its own Settings page, with Easy setup on Qualytics AI or Custom setup on your own provider.
-
Update Integration
Switch between Easy and Custom setup, or change the provider, model, key, or Business Context.
-
Remove Integration
Disconnect your AI provider from AgentQ.
-
Connect Azure OpenAI Through APIM
Point the Azure OpenAI provider at an APIM gateway, with the Base URL and header name that make it work.
-
Configure Access Controls
Pick how much AgentQ can see, with each level previewing the capabilities it turns on.
-
Select Dates
Change the period the audit reports on, beyond the last 30 calendar days it opens with.
-
Sort and Filter the History
Narrow the audit to the users you want to review, and order it by time, conversation, or user.
-
Start a New Chat
Create a fresh conversation from the sidebar, chat header, or floating chat.
-
Resume a Chat
Return to a previous conversation and continue from where you left off.
-
Rename a Chat
Give a conversation a descriptive title for easier navigation.
-
Archive a Chat
Move a conversation out of the active list into the archived section.
-
Restore a Chat
Bring an archived conversation back to the active list.
-
Delete a Chat
Permanently remove an archived conversation.
-
Search Chats
Filter active conversations by title or message content to find a specific conversation.
-
Troubleshooting
Symptoms, causes, and resolutions for setting AgentQ up, using it, and capabilities that are switched off.
-
API
REST API reference for AgentQ: chat, sessions, specialized endpoints, and configuration management.
-
FAQ
Answers to frequently asked questions about AgentQ, security, usage limits, and troubleshooting.
MCP¶
Connect an AI client you already use to Qualytics. This section covers how MCP works, the setup for each company's clients, best practices, and permissions, along with troubleshooting, the MCP API, and FAQ.
-
Getting Started with MCP
What the MCP server offers, and every page in its section.
-
Introduction
What MCP is, what the Qualytics MCP server offers, and why it matters for data quality.
-
How It Works
How a client, the Qualytics MCP server, and the model work together on a request, and what the model sees.
-
Connecting External AI Clients
What you need before connecting a client, and the setup for each company's clients.
-
Examples
Real scenarios built on the prompts the Qualytics MCP server ships with, from explaining an anomaly to judging whether data can be trusted.
-
Best Practices
Keep client connections secure and get reliable results from them.
-
Permissions
The user roles, tokens, and team permissions behind what a client can do.
-
Troubleshooting
Symptoms, causes, and resolutions for connecting a client, authorizing it, and calling tools.
-
API
The server endpoint, the authorization endpoints, scopes, and token lifetimes.
-
FAQ
Answers to common questions about the MCP server, supported clients, and how access is authorized.