E-commerce domain expertise for AI that has to get shopping right.
I evaluate, test and improve how AI systems handle products, catalogs and purchase decisions. Sixteen years in e-commerce, thirty-two in business, and hands-on work with AI models every day.

What I do
AI evaluation for e-commerce
I review and score AI outputs on real product data: shopping recommendations, product comparisons, catalog normalization and agent tasks such as online purchases. Rubrics with subscores and hard gates, labelled edge-case suites, documented error analysis.
price: wrong → FAIL (gate)
Learn more →Product data and catalogs
AI can only recommend what it can understand. I structure catalogs so models and shoppers read them the same way: category trees, attributes, variants, raw and canonical values. From focused premium ranges to one catalog run across five sales channels.
BLK → black
Learn more →AI visibility
I make brands appear when people ask ChatGPT, Perplexity or Google AI for a recommendation, and I measure it in GA4 against a baseline. My own store is the live test lab.
In numbers
How I work
- Define what good looks like. Before scoring anything, I write the rubric: what the user asked, what a correct answer must contain, and what fails it outright.
- Test the hard cases on purpose. Missing attributes, variants split into separate products, unit mismatches, budget conflicts, discontinued items.
- Slice the errors. I group failures by category, attribute and source, so the fix goes where the problem actually is.
- Document everything. Every judgement comes with the reason, the source and the date, so anyone can audit it.
Recognition




Need someone who knows both shopping and AI?
I work with AI teams, data platforms and e-commerce businesses. Remote, UK based.
Get in touch