The rationale layer

Every tool logs the what. None of them log the why.

Tacit Paradigm captures the reasoning behind decisions in the apps your team already uses: Stripe, Jira, Gmail, and Notion. It stays permanent, searchable, and available to people and AI agents.

runs locally · self-hostable · MCP-native
How it works

Capture, store, serve.

01

Capture

A Chrome extension and macOS app prompt inline where decisions happen. Silent on routine actions, it only asks when something is genuinely new.

02

Store

Every rationale becomes a node in a searchable decision graph. The reasoning stays even after the person who made the call leaves.

03

Serve

Past rationales surface back when they matter: to people in a dashboard, and to AI agents over MCP.

Integrations

Works where decisions happen.

Stripe
Jira
Gmail
Notion
Google Docs
Zendesk
Linearsoon
Salesforcesoon
For AI teams

Your decisions, agent-readable.

The rationale graph speaks Model Context Protocol. Point Claude Code, Cursor, or any MCP client at it and your agents act with the context of every decision your team has explained. Each capture is also a labelled (question, answer, context) pair, a fine-tuning dataset by construction.

Local inferenceScreen content never leaves the laptop.
Self-hostableGraph and MCP server run in your VPC.
# claude_desktop_config.json
{
  "mcpServers": {
    "tacit-paradigm": {
      "command": "tacit-mcp",
      "args": ["--workspace=acme"]
    }
  }
}

Stop losing the why.

See Tacit running on your tools in a 30-minute demo.

Book a demo