Artificial Intelligence pricing

We publish starting prices so you can budget before the first call. Every quote is fixed-fee: no hourly surprises, no scope creep surcharges.

Choose your engagement level

Three packages cover most scenarios. If your project falls between two, we will build a custom quote within 48 hours.

Assess

£2,400

One-time fee, two-week engagement

  • Audit of your existing data sources and quality
  • Feasibility report: which AI use cases are worth pursuing and which are not
  • Prioritised roadmap with estimated costs for each use case
  • One 90-minute workshop with your team to walk through findings
  • Written deliverable you can share with stakeholders or investors
Get started

Build

From £9,500

Fixed project fee, typically 4–10 weeks

  • Everything in Assess, plus full model development
  • Data pipeline engineering: cleaning, feature extraction, versioning
  • Model training, evaluation and hyperparameter tuning
  • REST API deployment to your cloud or ours
  • Monitoring dashboard with drift alerts
  • 90 days of post-launch support included
Get started

Managed AI

£1,800/month

Ongoing, minimum 6-month commitment

  • Continuous model monitoring and automatic retraining
  • Monthly performance report with accuracy, latency and cost metrics
  • Up to 20 hours of engineering time per month for improvements
  • Priority response: issues acknowledged within 4 hours on business days
  • Quarterly strategy call to align AI roadmap with your business goals
Get started

Feature comparison

A quick side-by-side view so you can see exactly what each plan includes.

Feature Assess Build Managed AI
Data quality audit✓✓✓
Feasibility report✓✓✓
Model development—✓✓
API deployment—✓✓
Monitoring dashboard—✓✓
Post-launch support—90 daysOngoing
Automatic retraining——✓
Monthly engineering hours——Up to 20 h
Quarterly strategy call——✓

How we price projects

Every Build quote is fixed-fee. We scope the work during the Assess phase (or a shorter scoping call if you already have a clear brief) and give you a number that does not change unless you change the scope. Here is what drives that number.

Data complexity. A single clean CSV is simpler than five databases plus a legacy ERP export. The more sources we need to integrate and clean, the more engineering hours go in.

Model type. A tabular regression model costs less than a custom object-detection model that needs thousands of labelled images. If labelling is needed, we quote it separately or train your team to do it with our annotation tool.

Deployment environment. Deploying to a managed cloud service like AWS SageMaker or Google Vertex is straightforward. Deploying to on-premise servers or edge devices (Jetson, Coral) requires additional testing and packaging work.

Compliance requirements. If your industry requires model explainability documentation, bias audits or GDPR data-processing impact assessments, we include those as line items so you can see exactly what you are paying for.

We do not charge by the hour for project work. Hourly billing rewards slow work; fixed fees reward efficiency. The only hourly element is the 20-hour monthly allocation in the Managed AI plan, and unused hours do not roll over.

Frequently asked questions

Can I start with Assess and upgrade to Build later?
Yes, and most clients do exactly that. The Assess fee is deducted from your Build quote if you proceed within 60 days, so you are not paying twice for the same discovery work.
What if my project is smaller than the Build starting price?
We occasionally take on micro-projects (a single model retrain, a data pipeline fix, a proof of concept) for less than £9,500. Tell us what you need and we will quote it honestly. If the scope is genuinely small, the price will be too.
Do you offer discounts for startups or non-profits?
We offer a 15% discount on the Build plan for registered charities and social enterprises. For early-stage startups, we can structure payment in two or three milestones instead of requiring the full fee upfront.
What happens if the model does not meet the agreed accuracy target?
We define a minimum performance threshold in the statement of work before any code is written. If we cannot meet it after two retraining cycles, we refund 50% of the Build fee. This has happened once in 47 projects, and in that case the data simply did not contain the signal the client hoped for. We identified that early and pivoted the project to a rules-based system that solved 80% of the problem at a fraction of the cost.
Can I cancel the Managed AI plan early?
After the initial six-month term, you can cancel with 30 days notice. During the first six months, early termination incurs a fee equal to two months of the plan. We keep it simple and put it in writing before you sign.