Learn via Chat — Documentation Answers in Conversation
- Ask "How do I...", "Can I...", or "What is..." about SRE Agent features — get answers from official docs
- No setup required — built-in and available on every agent
- Answers sourced from live documentation, so they stay current as features evolve
The problem: learning a new tool while using it
When you're setting up an agent or exploring what it can do, questions come up naturally: "How do I connect PagerDuty?" "What are run modes?" "Can I schedule recurring tasks?"
Without a documentation specialist, you have two options — both disruptive. You can leave the conversation, open the documentation website, search for the right page, read through it, and return to the chat with the answer. Or you can ask the agent directly and hope it knows the answer from its training data — but training data can be outdated or inaccurate for product-specific details like configuration steps, supported regions, or permission requirements.
Either way, you lose context. The investigation you were running, the setup you were configuring, the question chain you were following — all paused while you go find information.
How it works
Your agent has a built-in capability that activates automatically when you ask documentation-related questions. No configuration, no setup, no opt-in.
- You ask a question — type something like "How do I set up scheduled tasks?" in the chat
- Your agent recognizes the question type — questions about SRE Agent features, setup, or concepts trigger the documentation lookup
- It fetches the documentation index — reads the docs map listing all available documentation pages
- It reads relevant pages — identifies and fetches the specific documentation pages that answer your question
- You get an accurate answer — synthesized from official documentation, right in the chat
DocsGuide handles three categories of questions:
| Category | Examples |
|---|---|
| Features | "How does incident response work?", "What connectors are available?", "How do scheduled tasks work?" |
| Setup | "How do I connect PagerDuty?", "How do I create an agent?", "How do I set up a Teams bot?" |
| Concepts | "What are run modes?", "What is agent reasoning?", "What permissions does the agent need?" |
What makes this different
Answers come from official documentation, not training data. Your agent fetches documentation pages from sre.azure.com at runtime. When documentation is updated — new features added, steps changed, limits revised — your agent picks up the changes automatically.
No context switch. You stay in the conversation. The investigation thread keeps running. The configuration you were building stays open. The answer appears in the same chat window.
No setup required. This capability is built in, available on every agent from the moment it reaches Running state. No connectors to configure, no permissions to grant, no knowledge base to populate.
Before and after
| Before | After | |
|---|---|---|
| Finding setup steps | Leave chat → search docs → read page → return to chat | Ask in chat → cited answer in seconds |
| Answer accuracy | Agent answers from training data — may be outdated | Agent reads current, official documentation |
| Context preservation | Lost — you closed the tab, switched focus, forgot where you were | Maintained — answer appears in the same conversation |
| New feature discovery | Browse documentation site, hope to find relevant pages | Ask "What can this agent do?" and get a survey of capabilities |
| Time to answer | 2–5 minutes (searching, reading, returning) | Under 10 seconds |
Related capabilities
| Capability | What it adds |
|---|---|
| Custom agents → | How built-in and custom agents specialize your agent |
| Memory → | How your agent remembers context from past conversations |
| Knowledge Base → | Upload your own documentation for agent reference |
| Agent Playground → | Test this capability and others interactively |