Our standards

Our standards are set out in one place so you can check them before signing. The first part covers what every implementation includes; the second, the kind of systems we build when AI talks to people or decides about them. The links lead to the services where each principle is described in more depth.

What every implementation includes

• A pilot measured on your data. We agree the success criterion together before the start and measure it on your data, not on sample data. Each stage ends with a written result and your decision on the next step — set out in our delivery methodology.
• A person approves actions that have consequences. Anything that goes out to customers or changes data in your systems or how they operate starts in a mode where a person approves the actions. Loosening the oversight is a decision made on pilot data — described at AI agent implementation.
• The decision and the signature stay with the responsible person. In accounting, lending and cost estimating the system prepares the data and the working material — a proposed accounting entry, a quote, a cost estimate — and the person responsible for that domain, who knows the rules, the technology and the rates, makes the call. Professional responsibility stays where it belongs. We develop this at AI agent implementation and at AI in a construction company.
• Data protection, the GDPR and the AI Act, put in place with your lawyer. We prepare the register of AI systems, documentation, an activity log of what the system does (who, what, when) and a description of data flow. The controller and their lawyer settle the system’s classification, the legal basis and the impact assessment, and we implement a solution in line with that decision. Details at the AI Act technical compliance sprint and on the security and GDPR page.
• A processing location chosen for your requirements, not for the price on the quote. We compare the full cost: infrastructure, updates, backups and maintenance. We choose your own server when specific requirements justify it and someone is responsible for operating it. We agree the processing location and the boundaries of data flow and responsibility together before the start.
• Predictions built on data about past events. Route optimisation, failure prediction or scoring sales opportunities all need data covering enough past events — what counts is the number of recorded cases, not the number of months. If the company does not collect it yet, we start by putting the data collection in order and say plainly when the predictive result will arrive. How to check data readiness is set out under predictive maintenance; further examples at AI in a transport company.
• Documentation and a hand-over that leave you independent. The code, the configuration and the documentation are yours, and we hand them over at the end of the implementation. Your team learns to work with the system as part of the implementation — on the tools we have put live in your company, and on its own processes. More at AI implementation.
• Support with a clearly defined scope. We agree the supported systems, the work allowance and the response hours in writing. An automation can run around the clock, and we agree how support requests are handled with you before go-live. Our delivery methodology describes the hand-over.

The systems we build

• Every system is clearly labelled as AI. If a system is talking to a customer — by voice or in text — it is clearly marked as one, always. From 2 August 2026 Article 50 of the AI Act, the EU regulation on artificial intelligence, requires that disclosure too, although the article itself allows an exception where the machine is obvious from the circumstances and the context of use; we do not rely on it. Callers can ask for a person at any time: when an adviser is available, the bot transfers the call; outside staffed hours, it takes a message or a callback request — likewise our own delivery rule rather than a requirement of that article. Described at voicebot implementation.
• We build consented communication only. We automate contact with people who have asked for it or consented to it — for ourselves and for clients. AI-generated mass outreach to people who never asked for it is a legal risk — consent and electronic-communication rules — and a reputational one that campaign statistics do not show: a damaged reputation for the company’s e-mail domain, and for the company as a spam sender, outlasts the campaign. More at AI in a sales team.
• Voicebots call with offers only people who have agreed to it. Automated calling systems (machines that place calls on their own) may not send commercial information without prior consent — business recipients included — and you may not call in order to obtain that consent. The fine is imposed by the President of UKE and the exposure stays with the caller, which is to say with the client. Same reasoning as for bulk e-mail, with a penalty written directly into the Electronic Communications Law. More at voicebot implementation.
• In HR processes and customer service we work only with documents and the content of enquiries. Our systems have no biometric surveillance, no emotion recognition and no candidate assessments or rankings in recruitment — including ones a person would only approve. Inferring emotions in the workplace is prohibited by Article 5(1)(f) of the AI Act and has been since 2 February 2025. The prohibition covers recruitment and probation, binds the deployer (the company using the system) as well as the provider, and a candidate’s consent does not lift it. Set out at voicebot implementation and at the AI Act technical compliance sprint.
• A voicebot starts as a pilot on one call type. A voicebot (an automated system that holds phone conversations) starts on one call type, integrated with an existing system and with handover to a person; we widen the scope on the basis of pilot data. That way the automation proves itself in day-to-day work before it takes on more calls — described at voicebot implementation.
• The first voice project goes where callers are not under pressure. It covers routine, low-pressure service enquiries. We agree further applications on the pilot results, the caller's experience and the legal classification prepared by your lawyer. We develop this at voicebot implementation.
• Emergency calls are handled by people. Our voicebots handle business matters, not calls to emergency numbers. Evaluating and classifying emergency calls and prioritising dispatch is an Annex III high-risk use under the AI Act — the obligations for that category were deferred to 2 December 2027, which is deferred, not cancelled. We return to it at voicebot implementation.

Where we start

Every implementation starts by choosing a process where the investment in AI pays off; how we choose and measure it is set out in the delivery methodology. Most often we begin with a single document process — which one, and why, is on the AI implementation page, and how we automate processes step by step is shown at AI automation.

Frequently asked questions

What does every implementation include?

A pilot whose success criteria are agreed before the start and measured on your data; a system in which a person approves actions that affect customers or data; data protection put in place together with your lawyer; and the code, configuration and documentation, which we hand over at the end. Your team practises working with the system on its own tasks as part of the implementation, so after hand-over the company can develop it with us or on its own.

Why do you build only consented communication?

Because it protects your company in two ways. The first is legal safety: mass outreach requires consent, falls under electronic-communication rules, and the liability stays with the sender. The second is more practical: a damaged reputation for the company’s e-mail domain, and for the company as a spam sender, outlasts the campaign by a long way, and the cost of rebuilding it is not reflected in campaign statistics. So we automate contact with people who have asked for it or consented to it — in every project.

Can AI make decisions without a person involved?

It can, when the data supports it. Every implementation begins in a mode where the system prepares and a person approves. The data from that pilot — how many proposals were correct, what errors occurred and what they cost — shows which categories of action can proceed without approval. We take that decision together with you, on the numbers, after the pilot.

How do you handle compliance with the AI Act and the GDPR (General Data Protection Regulation)?

We work with your lawyer: they settle the legal classification, and we implement the solution. We prepare the technical and organisational side: the register of AI systems, documentation, an activity log of what the system does (who, what, when), human oversight, and the description of data flow that a legal assessment rests on. Classifying the system, choosing the legal basis for processing and running an impact assessment belong to the controller and their lawyer or data protection officer, and we build the system in line with their decision.