Knowing when to trust an AI model

A mid-sized UK market research company built digital twins using large language models to simulate how real consumers would answer surveys. The team believed in the potential but before committing further investment, the board needed evidence they could trust.

I designed an evaluation framework to test how closely the model’s simulated responses matched real survey data, then ran it against a representative set of cases to produce a score the board could act on.

Read the post →

Pricing used equipment with confidence

Dealers of used heavy machinery need prices they can defend to their customers. Setting prices requires years of experience and ongoing market research to keep up to date.

Equal Experts were approached by one of the world’s leading manufacturers of construction equipment to estimate used equipment prices by training a predictive model on past sales data.

Read the post →

Answering policy questions in seconds

The landscape for trade data is a confusing mesh of overlapping datasets. Policy-makers need to consult subject-matter experts to get answers to seemingly simple questions like “How many cars does the UK export to Germany”?

Working with Swirrl, Department for Trade and Investment (DIT) and the Office for National Statistics (ONS), I developed a search engine for trade data to help analysts find relevant facts to answer policy-makers’ questions immediately.

Read the post →

Data-driven consumer goods innovation

Vypr helps optimise products before they reach our supermarket shelves. Vypr gathers opinions from consumers to help retailers and manufacturers validate and develop fast moving consumer goods (FMCG). The platform provides a (disruptively) cost-effective way to get insights that save waste from launching new products destined for failure and tune innovation to improve profits.

Read the post →

Finding risks that spreadsheets miss

A large UK outsourcing company needed to know whether their financial forecast for a welfare-to-work programme could be trusted. Job seekers placed in work don’t always remain employed. Some cycle in and out of work many times over. The rare cases were uncommon, but disproportionately costly to resolve, and nobody could say whether the forecast was accounting for them properly.

Read the post →