Adam
An AI estimation copilot — customers describe a print job in plain language, Adam prices it in seconds.
The problem
Every quote request that waits overnight is a customer comparing competitors. Print estimation is genuinely fiddly — sizes, stocks, finishes, quantities — so it always queued behind a busy sales team. Hours of latency on a question the business could answer in seconds.
The system
Adam is a retrieval-augmented agent over the product knowledge base: paper stocks, finishes, machine constraints, pricing logic. A customer writes what they need in plain language — web or WhatsApp — and Adam parses specs, resolves ambiguity, and returns an accurate quote.
Conversions log automatically, closing the loop between enquiry and sale. Edge cases and empty rate-cards escalate to a human instead of guessing.
Deployed on Vercel as a production service in front of a real business.
Results
- Quote latency collapsed from hours to seconds on covered products.
- Runs in production — TODO: add quotes-per-month volume and the live demo URL (LINKS-TODO).