Artificial Intelligence FAQ

Most projects run between 8 and 14 weeks from kickoff to production deployment. The first two weeks are always a discovery phase where we assess your data, define success metrics and agree on the technical approach. After that, development happens in two-week sprints with a demo at the end of each one. A simple single-model project (for example, a churn predictor trained on one CRM export) can sometimes ship in six weeks. Larger multi-system deployments with integrations into ERP or warehouse management platforms tend to land closer to the 14-week mark.

Engagements typically start at £12,000 for a focused single-model project and range up to £60,000 or more for multi-system deployments with ongoing support. We quote a fixed price after the discovery phase, not a day-rate estimate that can balloon. The quote covers data preparation, model development, testing, deployment and a six-month support window. If the discovery phase reveals that the project is not feasible with your current data, we tell you that and charge only for the discovery work completed.

It depends entirely on the problem. A text classification model might work well with 3,000 labelled examples. A demand forecasting model for a retailer with 200 SKUs usually needs at least 18 months of daily sales history. During discovery we will tell you honestly whether your data is sufficient, and if it is not, we can help you design a collection strategy so you are ready in a few months rather than years. We have turned away projects where the data simply was not there, because shipping a model that guesses badly is worse than shipping nothing.

Your data stays in the UK. We use cloud infrastructure hosted in London-region data centres (AWS eu-west-2 or Azure UK South, depending on your preference). All data is encrypted at rest and in transit. We sign a data processing agreement before any data leaves your systems, and we delete all copies within 30 days of project completion unless you ask us to retain them for ongoing model retraining. If your compliance requirements demand on-premise processing, we can work with your own servers instead.

In our experience, no. The systems we build handle repetitive, high-volume tasks: scanning invoices, triaging support tickets, flagging anomalies in sensor data. That frees your team to spend time on work that requires judgement, creativity or relationship-building. One of our healthcare clients redeployed two full-time administrators from data entry into patient liaison roles after we automated their referral processing pipeline. Headcount stayed the same; job satisfaction went up.

Every project includes six months of support at no extra charge. During that period we monitor model performance, fix bugs and retrain the model once if your data distribution shifts significantly. After the six months, you can extend support on a monthly retainer (typically £800 to £1,500 per month depending on system complexity) or bring maintenance in-house. We provide full documentation, a runbook and a handover session for your internal team in either case.

Yes, integration is a core part of what we do. We have connected AI models to Salesforce, SAP, Microsoft Dynamics, Xero, custom PostgreSQL databases, SharePoint document libraries and several bespoke warehouse management systems. If your software has an API or a database we can query, we can plug into it. During discovery we map out every integration point and flag any that will need custom middleware.

Sometimes it is not. If a problem can be solved with a well-written SQL query or a simple rules engine, we will say so and save you the cost of a machine-learning project. AI tends to be the right tool when the rules are too complex or too numerous to write by hand, when the patterns in your data change over time, or when you need to process unstructured information like free-text, images or audio. Send us a short description of the problem and we will give you a candid opinion within a day.

We run half-day and full-day workshops on applied machine learning, aimed at developers and data analysts who want to understand how the models we build actually work. These are not theory lectures. Participants work through real datasets and build a small model from scratch by the end of the session. We also offer a two-hour "AI literacy" session for non-technical managers and directors, covering what AI can and cannot do, how to evaluate vendor claims, and how to set realistic project expectations. Workshop pricing starts at £1,800 for a group of up to ten people.