For me, Phase 2 should include not only the backend architecture, but also the business-level functionality needed to actually use the tool in real conditions:
Backend & database
PostgreSQL (source of truth)
ingestion, normalization, and persistence of all offers
duplicate handling at database level
full traceability (messages, files, suppliers)
Search & business logic
search APIs with filters (brand, product, size, packaging, supplier, date, price)
best price per product / packaging
latest offers
supplier-level analysis
Simple secured web interface
login access
product search with filters
best price display
latest offers view
product detail view (price history)
Manual entry
ability to add offers manually (phone / meetings)
Data correction
ability to edit extracted data
corrections reusable for future processing (basic learning)
Product normalization
handling naming variations (e.g. 70cl vs 700ml, naming differences)
Basic alerts
new best price
supplier becoming competitive
Show More