Artificial Intelligence FAQ
Below are the questions prospective clients ask most often. If yours is not here, ring us on +44 7789 846405 or email [email protected] and we will get back to you within a day.
Most engagements fall between £8,000 and £45,000. A focused predictive model built on well-structured data sits at the lower end. Multi-model pipelines with custom integrations, retraining automation, and role-based dashboards push costs higher. We always provide a fixed-price quote after the discovery sprint so there are no open-ended bills.
No. We handle model development, deployment, and monitoring end to end. Your side needs someone who understands the business problem well enough to validate outputs: a product manager, an operations lead, or even the founder. Technical handover documentation is included in every project if you later decide to bring AI skills in-house.
It depends on the use case. For a churn-prediction model we typically need 12 to 24 months of customer transaction records, support tickets, and any CRM fields you track. For NLP projects we need a representative sample of the documents you want to classify or extract from: emails, contracts, invoices, etc. During the discovery sprint we map exactly which fields matter and flag any gaps early.
A proof-of-concept model running on real data is usually ready within three weeks. Full production deployment, including API integration and monitoring, takes four to eight weeks depending on the complexity of your existing systems. We have completed simpler projects in as little as 18 working days.
Yes. All client data is stored on encrypted UK-based servers. We sign a data-processing agreement before any transfer takes place, and we delete raw data within 30 days of project completion unless you ask us to retain it for retraining. Our team members access data only through audited, role-restricted environments.
We have built integrations for Salesforce, HubSpot, SAP, Xero, and several bespoke ERPs. If your platform exposes a REST or GraphQL API, we can connect to it. For legacy systems without an API we have used database-level connectors or scheduled file exports as a bridge.
AI models degrade over time as the data they were trained on drifts away from current reality. We offer monthly retainer plans that include automated drift detection, retraining triggers, quarterly performance reports, and priority support. About 70 per cent of clients choose this option. The rest receive full documentation so their own teams can manage the model independently.
Our sweet spot is companies with 20 to 500 employees, but we have worked with solo founders and with divisions of FTSE 250 firms. The deciding factor is not company size but whether you have enough historical data to train a useful model. If you are unsure, send us a brief description of your data and we will tell you honestly whether AI is the right tool.
Understanding AI jargon
Clients sometimes feel lost in terminology during early conversations. Here is a short glossary of terms we use frequently so meetings feel less like lectures.
Model
A mathematical function trained on historical data to make predictions about new data. Think of it as a formula that learns its own coefficients from examples rather than being written by a programmer.
Training data
The historical records the model learns from. Quality matters more than quantity: 5,000 clean, well-labelled rows often outperform 500,000 messy ones.
Inference
The act of feeding new data into a trained model to get a prediction. When your CRM sends a customer record to our API and gets back a churn probability, that is inference.
Drift
When the patterns in live data diverge from the patterns the model learned during training. Seasonal changes, new product lines, or shifts in customer behaviour can all cause drift. Catching it early is why ongoing monitoring matters.
Still have questions?
We are happy to talk through your specific situation on a short call, no obligations attached.
Get in touch