You should be free to leave us.
We are not aiming to be the cheapest. We are aiming to be the ones after whom it works — and the surest sign of that is a client who stays because the service is useful, not because leaving has been made technically impossible.
What businesses like yours fear
These concerns are rational. Our job is to reduce them through architecture, commercial transparency and support.
Spending money without a clear return
Subscriptions accumulate, usage bills surprise, and nobody can say what each cost actually buys.
Systems your staff cannot understand
A clever system that only a specialist can operate is a liability, not an asset.
Confidential information leaving your control
Customer, driver, employee or case data quietly flowing into third-party services nobody agreed to.
AI mistakes reaching customers
A source record being turned into an invented sales claim, compliance conclusion or customer answer it cannot support.
Being locked in
Discovering that changing supplier means losing the system, the data, or both.
The system failing after launch
Something breaks at ten in the morning and there is nobody to call.
The current status of our registrations and insurance is listed item by item in the questions and answers.
Freedom to leave
Our objective is to build useful infrastructure on your side of the fence while keeping your dependence on us as low as the architecture allows. Where the agreed architecture permits, the accounts sit with you: you hold the hosting relationship, you pay the infrastructure provider directly, and the working system and its data remain available to you if our commercial relationship ends. What each component’s licence actually permits is written down for each engagement.
A recurring relationship should earn its place through monitoring, maintenance, incident response and ongoing development. We maximise your control and portability, and minimise artificial dependency through the design itself.
Your data stays yours
Every engagement documents where data originates, where it travels, where it is processed and stored, who controls each component, and which external providers receive anything at all. Where an external AI provider is proposed, you see why, and what flows to it. Where privacy, residency or control requirements justify it, suitable models can run privately — on infrastructure you control, or on UK or European infrastructure selected for the engagement. Privacy requirements shape the architecture, and the status of every claim we make about data handling is published for you to inspect.
Every pound should work
Price alone does not determine whether a model produces a better business result. Simple, repetitive steps run on simpler and cheaper approaches — sometimes ordinary code, rules or templates, with no AI involved at all. Premium models are reserved for work where premium capability earns its price. Before any component is recommended, it has to answer one question: what measurable value does this expense create for the business? We evaluate the full cost of ownership, including subscriptions, usage, integration, maintenance, support, migration and switching. This matters more than the headline price, and we actively resist the slow accumulation of paid services that each seemed small at the time. How that shows up in our own prices is published in one place.
Built to keep working
Your business should keep operating when one model, API, platform or integration becomes unavailable. Resilience is designed in from the beginning, with fallback depth following the criticality of the process — a critical workflow may route through an approved alternative model, fall back to deterministic handling, queue the work, or hand it to a person, so that a failure degrades gracefully instead of collapsing into an error. Every critical process is designed against one question: if the automation fails at ten tomorrow morning, how does your business continue operating? Monitoring is part of the product, because a system that fails silently is dangerous. Support after launch is a core phase of the work because models, vendors and APIs will keep changing after go-live.
We will tell you no
Professional restraint is part of what you are paying for. Each of these is a sentence we have written into our own method, and expect to say out loud.
Do not automate this.
Some processes are broken; automating them makes them worse faster. We say so before we build.
Keep the platform you already have.
If your existing software already solves the problem, the honest recommendation is to use it.
This subscription is unnecessary.
Before any new paid service: is the capability already present elsewhere, and what happens when the price rises?
Do not migrate yet.
Migration has a real cost. If the economics do not support it now, waiting is the professional answer.
This workflow needs human review.
Where risk requires it, a named person stays in the loop — designed in from the start, not bolted on.
This project is not a fit for us.
Restraint is part of the service. We would rather decline than deliver something we cannot stand behind.
Ask us the hard version of any of this
A fit call is the right place to test these commitments against your actual situation, including whether we would advise against automating it.