Send us an enquiry
21 Springfield Close, Manchester M1 5AN, Greater Manchester, United Kingdom
What to expect after you reach out
When your enquiry arrives, it goes to our project intake team. One of our senior consultants will read it, check whether we have relevant experience in your sector and reply within one working day. Here is what happens next.
If your project looks like a good fit, we schedule a 30-minute video call. On that call we ask about your data: what systems generate it, how much history you have, how clean it is, and who on your team manages it. We also ask what decision the AI needs to support and how you measure success today.
After the call, we send a short written summary (usually two or three pages) that outlines the problem as we understand it, the approach we would take, a rough timeline and an indicative cost range. There is no charge for this. If you decide to go ahead, we draw up a formal statement of work and schedule the discovery sprint.
We are straightforward about what AI can and cannot do. If we think your problem is better solved with a simpler approach, such as a rules engine or a well-designed dashboard, we will say so. Our goal is to give you the right tool, not to sell you the most expensive one.
Frequently asked questions
How much data do I need to start a project?
It depends on the problem. For tabular predictive models, a few thousand rows of clean historical records is often enough for a proof of concept. Document intelligence projects need 200 to 500 example documents. Image classification tasks typically require at least 1,000 labelled images per category. During the discovery sprint we audit what you have and tell you honestly whether it is sufficient.
Do you work with companies outside Manchester?
Yes. About half of our clients are based elsewhere in the UK, and we have worked with companies in Ireland and the Netherlands. Discovery calls and regular check-ins happen over video. For the discovery sprint we prefer at least one in-person workshop, which we can host at our office or at your premises.
What happens if the model does not perform well enough?
We define performance thresholds at the start of every project. If the proof of concept does not meet those thresholds, we investigate why: often the issue is data quality or labelling inconsistency, not the model architecture. We will propose a remediation plan. If the data simply cannot support the desired accuracy, we will tell you that and you will not be charged for further work beyond the agreed proof-of-concept fee.
Will I own the model you build?
Yes. All models, code and documentation we produce during an engagement belong to you once the project is paid in full. We retain no proprietary rights. We do ask for permission to describe the project in anonymised case studies, but that is optional.