EntityIntelliAI (Entity intelligence cache)

When to use

• Building or demoing agent knowledge CDN patterns: semantic cache + durable entity store + live retrieval.
• Wiring Tinyfish Search (GET https://api.search.tinyfish.ai) before fallback corpus.
• Pushing a research markdown artifact into Senso (POST /org/kb/raw) toward cited.md.

Stack (this repo)

• Next.js 15 App Router, Postgres (Ghost-compatible schema), Redis (semantic bundle), optional OpenAI.
• Tinyfish: TINYFISH_API_KEY; optional TINYFISH_SEARCH_URL, TINYFISH_SEARCH_LOCATION, TINYFISH_SEARCH_LANGUAGE.
• Senso: SENSO_API_KEY, SENSO_KB_FOLDER_NODE_ID, optional SENSO_API_BASE.

Key paths

• Pipeline: lib/researchPipeline.ts
• Tinyfish retriever: lib/researchRetriever/tinyfishRetriever.ts
• Senso ingest helper: lib/senso/ingestRaw.ts
• Publish API: POST /api/publish/senso with JSON { "markdown", "title?", "summary?", "query?" }
• Showcase checklist: docs/SHOWCASE.md

Commands

pnpm install
pnpm db:seed   # schema + demo entities (needs DATABASE_URL)
pnpm dev

Sponsor narrative (elevator)

Ghost + Redis keep entity truth; Tinyfish pulls live web evidence; Senso publishes grounded runs so agents on the open web can cite and builders can attach x402 / CDP / Stripe MPP / agentic.market at the fetch boundary.