Our story
Ai Impactful started in 2018 when our founder, Callum Greaves, left a senior machine-learning role at a London fintech to focus on a problem he kept seeing: companies spending six figures on AI projects that never left the prototype stage. The models were technically sound, but nobody had thought about how they would fit into existing workflows, who would maintain them, or what would happen when the data distribution shifted six months later.
He set up shop in Scotland with two engineers and a simple rule: every model we build must be running in production within 90 days, or we have failed. Seven years later that rule still holds. The team has grown to eleven people, the client list spans logistics, healthcare, retail and financial services, and 47 of our models are still live and monitored today.
We are based at 659 Cavendish Road, St. Torp, Scotland, XF2 7CC, United Kingdom, though most of our work happens remotely alongside client teams.
What we value
These are not slogans printed on a wall. They are the criteria we use when deciding whether to take on a project, how to staff it, and when to push back on a client request.
Production or nothing
A model in a Jupyter notebook is a research experiment. We measure success by whether the system runs reliably in production and delivers the business metric we agreed on at the start.
Honest scoping
If your problem is better solved with a SQL query or a rules engine, we will tell you. About one in five initial enquiries ends with us recommending a non-AI approach, and we think that is a sign of good judgement, not a lost sale.
Transparency in pricing
We quote fixed prices. You will know the total cost before we write a single line of code. If scope changes, we renegotiate openly rather than padding invoices with vague "additional research" line items.
Knowledge transfer
When the project ends, your team should understand what the model does, how to retrain it, and when to call us. We write documentation for engineers, not for other data scientists.
The team
Eleven people, mostly engineers, all based in the UK. Here are the four leads you are most likely to work with.
Callum Greaves
Founder and technical director
Priya Nair
Head of engineering
Tom Whitfield
Lead data scientist
Elena Marchetti
Project manager
Our timeline
2018
Founded in Edinburgh with three people and one client: a logistics firm that needed demand forecasting for perishable goods. That model is still running.
2019
Moved to St. Torp and expanded to five engineers. Delivered our first NLP project, a contract-review classifier for a mid-sized law firm.
2021
Reached 20 production models across eight clients. Added computer vision to our service list after hiring two specialists from the medical-imaging sector.
2023
Launched our ongoing support plans so clients could get monitoring, retraining and priority fixes without negotiating a new contract each time.
2025
Team of eleven, 47 live models, clients in logistics, healthcare, fintech and retail. Still operating under the same 90-day production rule.