What are Knowledge Agents?

Knowledge Agents are configurable AI agents that connect to your company's information sources and do knowledge work on your team's behalf. They answer questions, maintain content quality, run scheduled tasks, and integrate with the tools your team already uses—all from a single, governed place.

Every Guru workspace includes a default Knowledge Agent. Teams can also create specialized agents tailored to a specific department, use case, or audience—like a Support agent scoped to your help content, or a Sales agent connected to your CRM and competitive intelligence.

💡

Tip

Try it now: If you have Guru, you already have a Knowledge Agent. Just search for something in Guru - that's your default Knowledge Agent at work.

What Knowledge Agents can do

Answer questions. Ask a Knowledge Agent a question in plain language and it searches across your connected sources—Google Drive, Slack, Confluence, Salesforce, Guru Cards, and more—to return a verified, sourced answer. No manual searching across tabs.

Maintain content quality. Knowledge Agents don't just retrieve information—they help keep it accurate. Through Knowledge Agent quality, agents run nightly rules to automatically verify content that's being used and trusted, and unverify content that's outdated or flagged.

Run automations. Anything you can ask an agent in chat, you can schedule to run automatically. Knowledge Agent automations let you set up recurring prompts on a daily, weekly, or monthly cadence—so your agent can draft cards, send Slack updates, or call connected tools without anyone manually triggering it.

Connect to external tools. Through MCP (Model Context Protocol), Knowledge Agents can act as the knowledge layer for external AI tools—and access real-time data from connected platforms to include in answers and automated actions.

How Knowledge Agents are set up

Each Knowledge Agent has five configuration areas:

• Connected Sources. Choose which sources the agent can search—specific collections, external integrations, or all available sources.
• Permissions. Set who can use the agent and who has Owner-level access to manage its settings and view analytics.
• Behavior. Define the agent's tone, custom prompt, and what it says when no answer is found.
• Quality. Configure auto-verify and auto-unverify rules, set source limitations by verification status, and review the Quality Log.
• Skills. Enable additional capabilities like automations and content actions.

👥

Access Required

You must be an Owner on a Knowledge Agent to manage its settings. Workspace admins can manage all agents

Where Knowledge Agents appear

Once configured, a Knowledge Agent can surface answers in:

• Guru search bar — as the default answer experience
• Slack — attached to specific channels for suggested answers
• Browser extension — triggered on specific URLs or pages
• Specific Guru Pages — set a default agent per page
• External tools — via MCP or the Knowledge Agent API

Frequently asked questions about Knowledge Agents

Can I have more than one Knowledge Agent? Yes. You can create as many agents as your plan allows. Most teams start with one general-purpose agent and add specialized agents over time as they identify distinct use cases or audiences.

What's the difference between a Knowledge Agent and Guru search? Guru search returns a list of matching cards. A Knowledge Agent synthesizes an answer from across your connected sources—including external integrations—and cites the sources it used. It also learns from feedback to get more accurate over time.

Does a Knowledge Agent work on non-Guru content? Yes. Knowledge Agents can connect to external sources like Google Drive, Confluence, Slack, Salesforce, and more. For non-Guru sources, content must have been used by the agent in an answer before it's eligible for quality review.