How kumarmukesh88/ai-product-manager works
Codebasics Digital Asset Assistant indexes a local library of decks, PDFs, images, and public links so users can search its contents rather than filenames. Ingestion extracts document text and embedded images, applies OCR and text/visual embedding, and persists an index; search combines lexical, semantic, and visual evidence into ranked, confidence-gated results. Users can inspect match evidence, open source files, filter results, and reindex. The excerpts establish the API, indexing, retrieval, and UI connections; some extractor and scoring internals are unsampled, so uncertain links are omitted.
Main entry points: server.py HTTP API · ingest.py Index builder · search.py Hybrid search · app.js Search interface
Components 16
Ingestion
- Index builder
Asset Assistant/ingest.py - Content extraction
Asset Assistant/extractors.py - Search index
Asset Assistant/config.py
Retrieval
- Hybrid search
Asset Assistant/search.py - Confidence ranking
Asset Assistant/search.py
Application API
- HTTP API
Asset Assistant/server.py - Runtime configuration
Asset Assistant/config.py - Asset file service
Asset Assistant/server.py - Search feedback
Asset Assistant/server.py - Query event log
Asset Assistant/config.py
User experience
- Search interface
Asset Assistant/static/app.js
Other components
- Library user
- Local asset library
- Linked assets
- Embedding models
- OCR engine
Connections 26
- Library user → Search interface: searches
- Search interface → HTTP API: requests search
- HTTP API → Hybrid search: calls search
- Hybrid search → Search index: reads index
- Hybrid search → Embedding models: embeds query
- Hybrid search → Confidence ranking: scores candidates
- Confidence ranking → HTTP API: returns ranked results
- HTTP API → Search interface: returns results
- Local asset library → Index builder: supplies files
- Index builder → Content extraction: extracts assets
- Content extraction → OCR engine: reads image text
- Content extraction → Index builder: returns chunks
- Index builder → Embedding models: embeds content
- Index builder → Search index: writes index
- HTTP API → Index builder: rebuilds index
- HTTP API → Hybrid search: reloads index
- HTTP API → Search index: reads asset details
- Search interface → HTTP API: requests details
- Search interface → Asset file service: opens source or media
- Asset file service → Local asset library: serves original
- Library user → Search feedback: rates result
- Search feedback → Query event log: records event
- HTTP API → Runtime configuration: reads settings
- Index builder → Runtime configuration: reads paths and models
- Hybrid search → Runtime configuration: reads thresholds
- Local asset library → Linked assets: contains link references