AI that does a job, not a demo.

Models, chatbots and LLM features built on your data and measured against how you work today. If it doesn’t beat the baseline, we tell you before you spend more.

Book a scoping call

Measured against a baseline
Your data stays yours
Supported after launch

What we build

Four kinds of AI work

Most projects are one of these. Many combine two — a model that predicts, and an assistant that explains the prediction to staff.

Predict
Machine-learning models
Churn, demand, credit risk, pricing. Trained on your data, measured against the rule of thumb you use today, and only shipped if they beat it.
Converse
Chatbots & assistants
WhatsApp and web assistants that answer from your policies and product data, hand over to a person when unsure, and log every conversation.
Retrieve
LLM & RAG integration
Search and question-answering over contracts, reports and manuals — with citations back to the source page, so people can check the answer.
Operate
MLOps
Deployment, monitoring and retraining for models you already have. Drift alerts, versioned experiments and a rollback that works.

Honest advice

Where AI earns its cost — and where it doesn’t.

Good fits

Questions answered from long documents, with sources
Sorting, tagging and routing large volumes of text
Drafting first versions a person then checks
Forecasts where you have two or more years of history

Poor fits

Decisions that must be exactly right every time with no review
Problems a spreadsheet formula or a SQL query already solves
Predictions with little or no historical data
Anything where nobody will own it after launch

How a project runs

Small bets, then bigger ones

Scoping sprint
One week. We look at your data, agree the success metric and the baseline to beat, and give you a go/no-go with a fixed price.
Fixed fee, credited if you proceed
Prototype on your data
Two to four weeks. A working model or assistant, evaluated in front of you against the metric we agreed — not a slide deck.
Weekly demos
Production & handover
Deployed into your systems with monitoring, documentation and training for the people who will use it every day.
Then the support retainer

The monthly option

AI support retainer

Models drift, vendors change prices and deprecate versions, and users find edge cases. The retainer keeps what we built accurate, affordable and up to date.

Accuracy and drift monitoring
Model and prompt updates as vendors change
Monthly evaluation report
Priority fixes within one working day
5 hours of improvements each month
Cancel with 30 days’ notice

From$500per month

Chatbot and LLM upkeep for one production model or assistant. Additional systems at a reduced rate.

Book a scoping call

Questions

What clients ask before starting

Which models do you use?

Whichever is cheapest for the job at the required quality — often an open model you host, sometimes OpenAI, Anthropic or Google. We benchmark on your data and show you the trade-off before choosing.

Where does our data go?

It stays in your cloud wherever possible. When a hosted model is used, we pick enterprise terms where your data is not used for training, and we tell you exactly what leaves your systems.

What if the model doesn’t beat our current process?

Then we tell you at the end of the scoping sprint or the prototype, and you stop there. You pay for the stage you completed, nothing more.

Can you work with our in-house team?

Yes, and we prefer it. Pairing with your engineers is the fastest way for the work to outlast our involvement.

Next step

Bring the problem. We’ll bring the baseline.

A 20-minute scoping call to hear what you want to automate or predict, and whether AI is the right tool for it.

Book a scoping call

AI that does a job, not a demo.

Machine-learning models

Chatbots & assistants

LLM & RAG integration

MLOps

Good fits

Poor fits

Scoping sprint

Prototype on your data

Production & handover

AI support retainer

Bring the problem. We’ll bring the baseline.

AI that does a job, not a demo.

Machine-learning models

Chatbots & assistants

LLM & RAG integration

MLOps

Good fits

Poor fits

Scoping sprint

Prototype on your data

Production & handover

AI support retainer

Bring the problem. We’ll bring the baseline.