Back to all work
Case study · Agri price intelligence
Turning WhatsApp messages into price data
Vision LLM extraction of price information from WhatsApp messages, with validation rules and a human review queue.
Headline result
WhatsApp-to-data
- Industry
- Agri price intelligence
- Client
- TBD
- Result
- WhatsApp-to-data
The problem
Valuable price information arrived as WhatsApp messages and images, unstructured and hard to use.
What we built
A vision LLM extraction pipeline with Gemini as the primary model and Claude as the fallback.
Every extraction passes through 12 validation rules. Anything that fails goes to a human review queue instead of into the data.
Outcome
Messy WhatsApp messages become structured, validated price data.
Next case study
Get in touch
Have a similar problem to solve?
Talk directly to a senior engineer about what you're building.