Artificial Intelligence FAQ

These are the questions prospective clients ask most often. If yours is not listed, send us a message and we will answer within one working day.

Common questions

A focused proof-of-concept usually takes six to eight weeks from kick-off to a working prototype. That includes two weeks of discovery (data audit, problem definition, success metrics) and four to six weeks of modelling and API development.

Full production deployments range from three to six months depending on data complexity, integration requirements and the number of models involved. We give a timeline estimate at the end of the discovery sprint, once we have seen the actual data.

Discovery sprints are quoted at a flat fee, typically between £8,000 and £15,000 depending on scope. You get a feasibility report, a prototype, and a go/no-go recommendation at the end.

Full builds are billed on a time-and-materials basis with weekly burn-rate visibility. A mid-complexity project involving one model, one integration and standard monitoring usually lands between £40,000 and £90,000. Larger platform builds with multiple models and custom UIs can exceed that, but we scope them in phases so you can stop after any phase if the results do not justify continuing.

It depends on the use case. For predictive analytics we need historical transactional data, ideally at least 12 months. More is better, but we have built useful models on as little as six months of clean data.

For NLP projects we need a representative sample of the documents you want to process: 500 to 2,000 examples is a good starting point. We handle the labelling if you do not have labels already.

For computer vision we need images or video clips, along with any existing annotations. If you have no labels at all, we set up a labelling pipeline using tools like Label Studio and manage annotators on your behalf.

We sign an NDA and a data processing agreement before any data changes hands. Where possible, we work with anonymised or pseudonymised datasets. For clients in regulated industries (healthcare, finance), we can work entirely within your infrastructure so data never leaves your network.

We also support federated learning setups where the model trains on data that stays on your servers and only model updates are shared.

That is exactly what the discovery sprint is designed to catch. We define a minimum accuracy threshold at the start and test against it on held-out data. If the prototype does not meet the threshold, we explain why (usually insufficient data volume, poor label quality, or a problem that is genuinely hard for current methods) and recommend next steps.

Sometimes those next steps involve collecting more data or cleaning existing data. Other times the honest recommendation is to shelve the project. You only pay for the discovery sprint in that scenario.

Yes. We deploy models as containerised microservices (Docker, Kubernetes) that expose a REST or gRPC API. If your backend is Java, .NET, Node or anything else that can make an HTTP call, integration is straightforward.

We have integrated with Salesforce, SAP, Dynamics 365, Snowflake, BigQuery, PostgreSQL and MongoDB in past projects. If you use something less common, we will assess compatibility during discovery.

You do. All custom code, trained model weights and documentation are yours. We use open-source libraries (PyTorch, scikit-learn, Hugging Face Transformers) so you are never locked into a proprietary platform.

We retain the right to reuse general-purpose utilities and boilerplate code across projects, but anything specific to your data or business logic belongs to you exclusively.

We offer monthly support contracts that cover model retraining, performance monitoring, bug fixes and minor feature additions. Contracts are month-to-month with 30 days notice to cancel.

If you prefer to bring support in-house, we run a knowledge transfer session at the end of the build phase and hand over full documentation, including a runbook for retraining and deployment.

Our office is at 72 Abshire Chase, London EC1V 9NR, Greater London, United Kingdom. We are a short walk from Old Street station. Meetings can be in person or over video call, whichever you prefer.

Phone: +44 20 1035 8421. Email: [email protected].

Still have questions?

We are happy to jump on a 20-minute call to discuss your situation. No pitch, no pressure. If AI is not the right tool for your problem, we will tell you.

Send us a message