A small team with deep technical skills and a clear focus: build AI tools that pay for themselves quickly.
Ai Essential Tools started in 2019 when our founder, Rhian Carver, left a data science role at a Cardiff insurance firm. She had spent three years watching promising AI prototypes die in PowerPoint because nobody connected them to the systems people actually used every day. The gap between what machine learning could do in a Jupyter notebook and what it did in production was enormous.
So she started a consultancy with a single rule: every project must be in production within 90 days or it gets cancelled. That constraint forced us to pick problems carefully, keep scope tight and involve end users from day one.
Six years on, the team has grown to eight people. We still follow that 90-day rule, although most projects ship well before the deadline. Our office sits on Dach-Kilback Glade in Towneing, a quiet spot that suits the focused, heads-down work we do. Clients visit occasionally, but most collaboration happens over video calls and shared dashboards.
We have turned down more projects than we have accepted. If a problem is better solved with a spreadsheet formula or a simple database query, we say so. That honesty has earned repeat business from clients who came back months later with a genuinely suitable AI challenge.
These are not slogans on a wall. They are the rules we use when making decisions about which projects to take, how to scope them and when to push back on a client request.
We agree on a baseline metric with every client before writing code. If we cannot define what success looks like in numbers, we do not start. This keeps projects honest and gives both sides a clear way to evaluate the result.
Large AI rollouts fail more often than small ones. We deploy a minimal version to a pilot group within the first few weeks, gather feedback, fix what is wrong and expand from there. The pilot group usually consists of three to five users who represent the full range of skill levels on the team.
We do not hand over a model and disappear. Our support plans include monthly performance reviews where we check whether accuracy has drifted, whether usage patterns have changed and whether the tool still saves the time it was designed to save. If it does not, we retrain or redesign at no extra cost during the support period.
Client data stays on client infrastructure wherever possible. When cloud processing is necessary, we use UK-based data centres with encryption at rest and in transit. We never use client data to train models for other clients, and we delete all copies within 30 days of project completion unless you ask us to retain them for ongoing support.
We are a team of eight. Everyone writes code, everyone talks to clients. There is no layer of project managers between the people who build the software and the people who use it.
Founder and lead data scientist
Rhian spent seven years in insurance analytics before starting the company. She specialises in NLP and document understanding models. When she is not reviewing pull requests, she runs a local coding workshop for teenagers.
Senior engineer
Tom handles infrastructure and deployment. He has built CI/CD pipelines for machine learning models at three different companies and joined us in 2021. His background is in DevOps, which means our models actually run reliably in production rather than just passing tests in development.
Machine learning engineer
Seren joined straight from a PhD in computational linguistics at Swansea University. She builds the classification and extraction pipelines that power our document intelligence product. She is also the person who insists on writing thorough test suites, which the rest of us quietly appreciate.
A short history of the company so far.
Rhian registers Ai Essential Tools as a sole trader and completes the first project: an invoice classifier for a small accounting practice in Swansea. The tool processed 400 invoices per day with 94% accuracy on the first version.
Hired Tom Elias as the first employee. Moved from a home office to a shared workspace. Completed six projects that year, mostly document automation for financial services firms.
Grew to five team members. Launched the predictive analytics offering after a successful pilot with a logistics company in Cardiff. Revenue crossed £300,000 for the first time.
Moved into our current office on Dach-Kilback Glade. Reached 42 completed projects across 28 different clients. Started offering ongoing support plans, which now account for about a third of revenue.
Team of eight. Currently developing a workflow automation toolkit that will let smaller businesses configure AI decision rules without writing code. Expected launch in Q3.