Investigate
Ask the Sazabi agent about your systems in a thread, work across surfaces, use the agent's tools and sandbox, and lean on project memory.
Investigation in Sazabi happens in a thread: you ask a question in plain language and the Sazabi agent searches your data and answers with the evidence it found. This group covers how to run those investigations well — threads, how to work with the agent, the tools it uses, the sandbox it runs commands in, project memory, and how to chat from Slack or Microsoft Teams.
In this section
Threads
Start a thread and keep it in sync across surfaces.
Working with the agent
Steer investigations and read the artifacts the agent produces.
Agent tools
Log search, web search, code, issues, and integrations.
Confidence & limits
What the agent can and cannot conclude from your data.
Set up the sandbox
Give the agent CLIs to run against your systems.
Use project memory
What the agent remembers and how to curate it.
Chat outside the dashboard
Ask the agent from Slack or Microsoft Teams.
Troubleshooting investigations
Fix common investigation problems.