What is daimon?
daimon is an open-source data science agent that answers analysis questions inside Discord, Slack, or Microsoft Teams threads. Mention it to have it write and run code, fit Bayesian models with PyMC, and return charts, an explanation of uncertainty, and a notebook that reproduces the result. It is designed for teams that want to discuss data in their existing workspaces, with shared thread context and options for recurring analyses. Organizations self-host it and provide an API key for the agent provider.
daimon Features:
Thread-Based Analysis: Mention daimon in a channel to start an analysis in a thread, where it writes and runs code and posts its findings.
Models and Charts: Analysis replies can include fitted Bayesian models using PyMC, charts, and a plain-English explanation of uncertainty.
Runnable Notebooks: Results can include a marimo notebook that reproduces the analysis, so the work can be inspected and run again.
Shared Thread Context: Team members can follow up in the same thread, and sessions retain context across turns.
Scheduled Routines: Recurring analyses can run without a prompt and post their results to a channel you choose.
Tool Connections: daimon can connect to Notion, Linear, other MCP servers, and GitHub repositories through its GitHub App.
Multi-Workspace Hosting: One self-hosted deployment can serve multiple Discord servers and Slack workspaces as isolated tenants; a Teams deployment serves one Microsoft 365 organization.
daimon Benefits:
Clearer Decisions: An explicit read on uncertainty helps teams distinguish a meaningful result from one that may be noise.
Easier Verification: A reproducible notebook gives analysts a way to inspect how a result was produced and rerun the work.
Shared Understanding: Keeping questions and follow-ups together makes analysis easier for the people in the thread to review and discuss.
Less Repeated Work: Scheduled analyses can deliver recurring updates without someone having to start each run manually.
Separated Communities: Tenant-level isolation lets multiple workspaces use one deployment while keeping each community's agents, memory, and data separate.
Use Cases:
Pricing Review: Analyze sales data around a pricing change to assess whether the change appears to have helped.
Experiment Comparison: Compare A/B test variants and assess whether the observed difference is likely to be more than noise.
Signup Forecasting: Forecast next month's signups and share uncertainty bands with a team in its chat workspace.
Weekly Metrics: Schedule a Monday summary of weekend numbers to post in the channel where the team reviews them.
Collaborative Analysis: Let teammates ask follow-up questions about a chart or model in the same conversation thread.

