The Challenge
A second-hand goods retailer had built up a large database of products they had successfully sold over the years. That history was full of valuable signal, which categories sold fastest, at what price, and at what margin, but it was sitting idle, locked away in raw sales records.
To keep the business growing, someone had to manually research wholesale and reseller marketplaces, hunt for items worth restocking, guess at likely resale value, and weigh up whether the margins justified the purchase. It was slow, repetitive, and entirely dependent on one person’s time and intuition.
The opportunity was clear: the answers were already hidden in the client’s own data. The challenge was building a system intelligent enough to read that history and turn it into sourcing decisions automatically.
The Solution
I designed and developed a bespoke AI sourcing engine that connects the client’s historical sales data directly to live wholesale marketplaces, doing the research a person would otherwise spend hours on, in seconds.
Rather than a generic product search, the system reasons from the client’s own results. It understands what they actually sell, what it sells for, and where the best return on investment is most likely to come from.
Learning From Real Sales Data
The system ingests the client’s full sales history and uses AI to make sense of it, recognising patterns across thousands of past sales rather than relying on guesswork. From this it builds an accurate picture of:
- Which product categories perform best.
- Typical resale prices and achievable margins.
- What is worth sourcing again, and what isn’t.
Automated Categorisation & Sourcing
AI automatically categorises every product line, then searches global wholesale and reseller marketplaces for comparable items. Each candidate is scored on projected return on investment, so the most profitable opportunities rise straight to the top:
- Automatic categorisation of the entire product range.
- Live search across wholesale marketplaces.
- ROI and margin ranking based on the client’s real results.
- A prioritised shortlist instead of an endless list to sift through.
Side-by-Side Visual Verification
Numbers alone don’t confirm a match. The system uses AI vision to compare a sourced product against the client’s previously sold item side by side, helping confirm it really is the same, or a better, product before any money is spent.
The Result
What was once hours of manual research is now an automated, continuously running process. Instead of trawling marketplaces and estimating margins by hand, the client receives a ranked shortlist of the highest-ROI sourcing opportunities, drawn directly from their own proven sales history.
The system removed a significant amount of effort and research from the business while making sourcing decisions sharper, faster, and grounded in real data.
Benefits Delivered
- Hours of manual marketplace research eliminated.
- Sourcing decisions driven by the client’s own sales data.
- Automatic categorisation of the entire product range.
- Opportunities ranked by projected ROI and margin.
- AI vision verification before purchase.
- A repeatable system that keeps working in the background.
Why This Project Matters
This project is a clear example of what AI now makes possible for everyday businesses, not as a novelty, but as a genuine competitive advantage.
Most businesses are sitting on data they have never been able to fully use. By combining that data with modern AI, it becomes possible to automate research, surface opportunities, and make better decisions faster than any manual process ever could.
My goal is to solve real problems for clients. So much of AI’s potential is still untapped, most people only ever use it for simple questions. This project shows what happens when you point it at a real business problem instead: it quietly takes on the heavy lifting and gives the business owner their time, and their edge, back.
Technology Stack
Frontend
- Next.js 14 (App Router)
- React 18
- TypeScript
- Tailwind CSS
Artificial Intelligence
- Claude (Anthropic) for categorisation & reasoning
- AI vision for side-by-side product verification
- Sales-history analysis & ROI scoring
Data & Sourcing
- SQLite product & sales database
- Automated wholesale marketplace search integration
- Historical sales ingestion pipeline