What is an AI agent, in plain English?
Software that does a job rather than answering a question. It watches for something happening, gathers what it needs from your systems, works out what should happen next, does it, and brings anything above its limits to a person. Think of it as a very fast, very literal member of staff who never forgets a step.
Is this just a chatbot with a different name?
No. A chatbot holds a conversation. An agent finishes work. Most of the agents we build never speak to a customer at all: they read orders, build quotes, chase invoices, prepare reports and update systems. We do build chatbots when a business needs one, but that is a different product.
How is an agent different from an automation like Zapier?
An automation follows the path you drew, in the same order, every time. It is excellent until something turns up that the path did not allow for. An agent works out the path, which means it copes with the exceptions that make up most of real business. We use both, and part of the blueprint is deciding which each job needs, because a simple automation is often the better answer.
Will it work with the software we already use?
Almost certainly. We work with the main CRMs, accounts packages, ecommerce platforms, inboxes and reporting tools, and if a system has an API we can usually reach it. We do not ask you to move platform. The whole point is making what you already run smarter.
What actually happens to our data?
Everything runs on your own accounts, so your data stays in your systems. Access is scoped to what each agent needs, every action is logged, and the processing and model arrangements are set out in writing before anything touches live data. We go through all of it during the blueprint rather than after you have signed.
Are we replacing people with this?
Not in our experience. What gets replaced is the copying and pasting, the rekeying and the chasing, which is usually the part of the job people like least. The businesses we do this for tend to grow without hiring for admin, and put the people they have onto work that needs judgement.
How quickly will we see something actually working?
Your first agent is live inside month one on both plans, doing a real job somebody does today. The blueprint comes back within a fortnight of starting, and production systems across a department usually take 60 to 90 days.
Do we need technical people on our side?
No. You need people who know how the business works, which is who we sit with during the blueprint. We handle the building, the integrating, the testing and the training, and everything is documented so your team can use it without understanding how it was made.
What does it cost?
Ongoing work is £2,500 a month part-time or £4,950 full-time, excluding VAT, with a three-month minimum. The blueprint is £2,000 and is credited in full against a six-month retainer if you go ahead within 60 days. Model and token costs sit on your own account and are quoted up front.
Do we own what you build?
Yes, entirely. The agents, the workflows, the software, the prompts and the documentation are yours. If you stop working with us you keep all of it and we hand it over properly rather than switching it off.
What happens when an agent gets something wrong?
It should not get far, because anything above the limits you set stops and waits for a person, and every action is logged so we can see exactly what happened. When something does slip through, we fix the cause rather than the symptom, and that is a large part of what the ongoing work pays for.
What if we want to stop?
The retainer has a three-month minimum and is then rolling, cancellable with 30 days notice. You keep everything built, fully documented, and we hand it over properly rather than switching it off.