An AI agent is a piece of software that does a job on a computer the way a person does it: it opens the system, reads what is there, makes a judgement, and finishes the task. Ours are built for the repetitive work that eats hours in a business — and then looked after monthly, because anything that runs your business needs somebody whose job it is when it breaks.
What an agent actually is
There is a great deal of noise about artificial intelligence, and very little of it is useful to a business owner. So here is the plain version.
A chatbot answers questions. An agent does work. Given the right access, ours will log into a supplier portal, pull a customer’s details across into your system, draft the follow-up, file the document, reconcile the list, or answer the enquiry that came in at ten past nine on a Sunday. It runs continuously, it does not get bored, and it produces a written record of what it did.
What it is not is magic, and it is not a replacement for judgement. Anything with a person’s livelihood or legal position riding on it goes to a human — our job is to make sure the human only sees the cases that genuinely need one.
What we automate
The starting point is never the technology. It is a conversation about where the hours go — what your team does every single day, in the same order, that is necessary but has stopped being interesting. Typical work we take on:
- Data movement. Pulling information out of one system and into another, including the badly-behaved ones with no export button.
- Enquiry handling. Reading incoming email, text and web forms; answering the routine ones; escalating the ones that are not routine; chasing the ones that went quiet.
- Document work. Extracting the same fields out of the same forms, checking them against a record, and flagging what disagrees.
- Reporting. Assembling the weekly or monthly numbers from the places they actually live, and emailing them to the person who asks for them.
- Monitoring. Watching the things nobody remembers to check — stock levels, listings, broken pages, a spreadsheet that stopped updating.
- Follow-up. The quoting, booking and chasing that everyone intends to do and nobody has time for.
If a process is written down anywhere — a procedure document, a checklist, a set of steps in somebody’s head that they can describe — it can usually be automated. Businesses that already document their processes are the easiest to work with, because a human following a written procedure to the letter is already close to a specification.
What it costs
We publish our prices. A service where the answer is always “let us have a call” is not one you can plan around.
- Agent audit — free. A short conversation and a written, ranked list of what is worth automating in your business, with the effort and the likely value against each item. Take it away and act on it yourself if you prefer. If you go ahead, it is credited against the build.
- Your first agent — £1,450. One workflow, live within seven days. The agent itself is running within twenty-four hours so you can watch it work, and thirty days of tuning are included.
- Care — £295 per month. One agent looked after: monitoring, fixes, a short weekly note showing what it did, and one new capability a month added at your request.
- Operations — £495 per month. Up to three workflows, a monthly strategy call, and a priority response when something needs attention.
- Fleet — £950 per month. Several agents across several systems, a fixed weekly slot, and dedicated reporting.
All prices exclude VAT. There is no metering: not per token, not per user, not per run, and no charge for the volume of work an agent gets through. What is capped is scope — the number of workflows we look after — because that is the part that actually costs us time. We would rather say that plainly than sell you a cheap plan with a usage meter hidden behind it.
How it works
- The audit. We find the work worth doing and agree which single job comes first — the one with the least effort and the most value, never the most impressive one.
- Alive in a day. Your agent is running within twenty-four hours, on your own material rather than a demonstration.
- Working within a week. The first automation is solved end to end inside seven days. Where a written procedure exists we follow it; where it does not, we write it down and you confirm it before anything is automated.
- Looked after, weekly. Every week a short note with the numbers before and after. Every month, one more capability. Every quarter, we revisit what else is worth doing, because scratching one itch tends to reveal the next.
What we will not automate
Some things should not be touched by an agent, and a supplier who will not say so is not one to trust.
- Regulated wording. Financial promotions, credit, lending and insurance wording are never written or rewritten by an agent. Where your business is regulated, the agent routes that work to a human, every time.
- Clinical, medical and care records. Patient data does not go through an agent we run. It stays where it belongs.
- Anything irreversible without a human. Payments, deletions, contract acceptance and anything a customer could be harmed by are gated behind an approval step. The agent prepares the work; a person commits it.
- Impersonation. Our agents have their own names and their own email addresses. They do not pretend to be you, and they do not sign your name to anything.
Your data and your accounts
An agent that does your work sees your information, so the arrangement has to be explicit.
- Where we handle personal data on your behalf we do it as a processor under UK GDPR, in a written agreement, and only for the purpose you have asked for. Data protection is built into the scope, not bolted on afterwards.
- The agent runs on your accounts, in your tenancy, so the access lives with you. We do not keep copies of your logins.
- Credentials are held in an encrypted store, never in email, chat or a document.
- If you leave, you keep the accounts, the access and a written runbook covering how everything works and how to switch it off. A service you cannot leave is not a service.
Why us
We have run this technology in our own business for long enough to know where it breaks, what it costs to run, and which tasks are a waste of everybody’s time. Our own agency runs on it: the reporting, the monitoring, the research, the routine admin. We are not reselling somebody else’s platform or passing on a per-seat licence, which is precisely why the running cost is small and why we can quote a flat fee. Built on open-source agent software, running on infrastructure we manage ourselves.
Questions we get asked
Do we need to replace our systems?
No. The point of an agent is that it works with the tools you already use — including the ones that are not very good. If a system is genuinely unusable, we will tell you, and quoting its replacement is a separate conversation.
How long before it is doing something useful?
The agent is running within a day, the first automation is working within seven. What takes longer than people expect is agreeing which job comes first — and doing that properly is what the audit is for.
What happens when it gets something wrong?
It will, occasionally, and the design assumes it. Work that matters is checked before it commits, everything is logged so we can see what happened, and fixes are covered by the monthly fee. You get a named person to tell rather than a ticket queue.
Is this only for large companies?
No. The businesses that benefit most are usually the ones with five to fifty staff, where a single repetitive job is eating a day a week. The entry point is £295 a month precisely so that starting is not a leap.
Can I cancel?
Yes, monthly, with the accounts and the runbook. If the service is only worth having because leaving is difficult, it is not worth having.
Start with the free audit.
Tell us the job that eats your week. We will tell you honestly whether an agent is the right answer for it — and if it is not, we will say so.
