Rhiannon Hale
Founder and lead data scientist
Eight years in insurance analytics before starting Fair AI Ventures. Builds most of our predictive models and handles client-facing technical discussions.
Our story
Fair AI Ventures started in late 2021, when our founder Rhiannon Hale left a data-science role at a large Cardiff insurer. She had spent three years building models that worked well in notebooks but rarely made it into production. The gap between a promising prototype and a system that runs reliably every Monday morning was enormous, and most consulting firms she encountered either ignored that gap or charged six figures to bridge it.
She registered the company with a simple thesis: small and mid-sized businesses in Wales and the wider UK deserve access to the same AI capabilities that large enterprises use, at a price that makes commercial sense. The first project was a churn-prediction model for a 30-seat SaaS company in Swansea. It took five weeks and cost £9,500. That client is still with us.
Today the team has grown to seven people. We operate from a small office in East Mertz and work remotely with clients across the country. We have delivered 58 projects since launch, and our repeat-engagement rate sits at 64%. That number matters more to us than any award or partnership badge.
What guides our work
Four principles we apply to every project, not just the ones with big budgets.
-
Transparency over mystique
We explain every model in plain language. If we cannot describe why a prediction was made, we do not ship it. Clients receive documentation that a non-technical manager can read and challenge. Black-box outputs erode trust, and trust is the only reason anyone keeps using a model after the novelty wears off.
-
Fixed pricing, no scope creep
Every project gets a written quote with a defined scope. If we underestimate the effort, that is our problem, not yours. If you request changes mid-project we quote those separately before starting. Two of our competitors lost clients to us specifically because of surprise invoices. We would rather earn less on a project than damage the relationship.
-
Data stays in the UK
We process and store all client data on UK-based servers. No data leaves the country unless you explicitly ask for it and sign a separate agreement. For regulated industries like healthcare and finance, this is not a nice-to-have. It is a requirement, and we treat it as one from day one.
-
Honest feasibility assessments
We turn down roughly one in five enquiries because the data is insufficient or the expected ROI does not justify the cost. Saying no early saves everyone time and money. When we do say no, we explain what would need to change for the project to become viable later.
The team
Seven people, each with a specific role. No one carries the title "AI guru."
Tom Bevan
ML engineer
Handles model training pipelines, infrastructure, and deployment. Previously built recommendation systems at a London e-commerce company for four years.
Sian Pritchard
Project manager
Keeps timelines honest and clients informed. Ten years of project management experience across software and construction before joining the team in 2022.
Amir Osei
Backend developer
Writes the integration code that connects our models to client systems. Specialises in REST APIs, message queues, and database optimisation.
Three additional team members work part-time on data annotation, quality assurance, and administrative support.
By the numbers
A snapshot of where we stand as of early 2025.
58
Projects delivered
64%
Repeat-engagement rate
7
Team members
90
Days of free post-launch support