Building Your Own
Knowledge Graph for AI

This brief creates the foundation — a Google Sheet with 4 tabs that follows the Fodda Pattern Standard:

• Signals — Raw data: articles, observations, bookmarks. 13 columns including title, summary, brands, sector, and pattern_name
• Patterns — Your analysis: recurring behaviors and emerging trends. Each supported by 2+ signals
• Entities — Tags: brands, technologies, audiences, locations that appear across your signals
• Graph Meta — Identity: who you are, what your graph covers, your editorial perspective

This brief builds a local web app with two core features:

• Signal Collection — Paste a URL → the app fetches the article, extracts metadata via LLM (summary, brands, sector), and writes it to your Sheet
• Pattern Recognition — Select untagged signals and either create patterns manually or use AI to suggest groupings

Tech stack: Vite + React frontend, Node/Express backend, Google Sheets API, Gemini API for extraction.

7 API endpoints. Full LLM prompts included. Auto-populates the Entities tab.

Once you have signals and patterns in your Sheet, Fodda makes it queryable by AI:

• Step 1: Share your Sheet read-only with Fodda's service account
• Step 2: Register at app.fodda.ai
• Step 3: Fodda validates your Sheet structure
• Step 4: You receive an MCP API key — the URL you use to connect Claude, Notion, or any AI tool
• Step 5: Confirm your dataset info (description & perspective are most important)

Your data stays in your Google Sheet. Fodda only reads it.

If you want custom query logic, caching, or a data source beyond Google Sheets — build your own MCP server. Not required if you're using Fodda's hosted path.