Frequently asked questions about Artificial Intelligence

Most engagements run between four and twelve weeks. A simple predictive model built on clean, tabular data can reach production in four weeks. Projects that require extensive data cleaning, custom labelling or integration with legacy systems tend to land closer to twelve. We give you a week-by-week timeline during the scoping phase so there are no surprises.

That depends entirely on the problem. For demand forecasting we typically need at least 18 months of transaction records. For a document classifier we need a representative sample of 500+ labelled documents. We sign a mutual NDA before any data changes hands, and we can work inside your cloud environment if you prefer the data never leaves your infrastructure.

No. We handle the full pipeline: data preparation, model training, deployment and monitoring. What we do ask for is a domain expert on your side, someone who understands the business problem well enough to validate whether the model outputs make sense. A product manager or operations lead usually fills that role well.

Our smallest fixed-price engagements start at around £8,000 for a proof-of-concept model with a single endpoint. Full production systems with monitoring, retraining pipelines and SLA support typically fall between £25,000 and £80,000. We quote fixed prices, not day rates, so your budget is predictable from day one. The scoping call is free and takes about 45 minutes.

Every deployment includes a 90-day monitoring window at no extra charge. During that period we track accuracy, latency and data drift. If the model degrades beyond the threshold we agreed in the contract, we retrain it at our cost. After the 90 days you can either take full ownership of the system or move to a monthly support plan that includes monitoring, retraining and priority bug fixes.

Yes. We deploy on AWS, Google Cloud and Azure regularly. If you already have a Kubernetes cluster or a managed ML platform like SageMaker or Vertex AI, we will use it. We can also deploy to on-premises servers or edge devices if your data governance rules require it.

We treat every client dataset as confidential by default. All transfers happen over encrypted channels, and we delete your data from our systems within 30 days of project completion unless you ask us to retain it for ongoing support. We are registered with the ICO and comply with UK GDPR. If your organisation has specific security requirements, we are happy to complete your vendor assessment questionnaire before work begins.

During the scoping phase we define a minimum accuracy or performance threshold together. If the model fails to meet that threshold after the first training cycle, we investigate why: sometimes the data needs more cleaning, sometimes the feature set needs rethinking. We include one full retraining round in every contract at no additional charge. If after that second attempt the model still falls short, we will tell you honestly and you will not be invoiced for the training work.