Floor plans to asset register
An Irish facilities company
26,852 door assets across 97 buildings in about eight weeks
Every AI product has a human layer. The question is who runs it.
We run the human layer behind your product as a team managed from London: tech-literate operators who follow your spec, with bots that take the repetitive work.
Your engineers should not be doing this.
Behind almost every tech product there is a layer of manual work: data to label, clean and check; AI output to review; edge cases; customers who need a real answer; onboarding checks.
Engineers end up doing it, which is the most expensive way to do it.
Your engineers are doing data cleaning. That’s a very expensive data cleaner.
What we do
For founders, COOs and heads of operations, support or data at SaaS, AI, marketplace, fintech and e-commerce platform companies.
- Human-in-the-loop review of AI output.
- Data operations: cleaning, structuring, labelling and validation.
- QA: quality checks against a written specification.
- First- and second-line support from people who can use the product.
- Customer operations and onboarding checks.
- Documenting manual processes so parts can be automated later.
Data labelling is a capability. We have no reference project for it yet.
Human in the loop is not a weakness in your product. It is the quality system.
How we work
Inside your stack from day one
We work in your own tools: Slack, helpdesks, order systems and custom internal tools.
We follow your spec
We work to your specification and check quality against it. Every task is logged: what job you gave us, and when we did it.
Cross-checking
On data projects, operators work in parallel and every file is cross-checked.
Weekly batches and reports
On data projects, work arrives in weekly batches, with a progress report every week. Every week we also send a report against the service standards (the SLA) we agree with you.
Edge cases come to you
Anything outside the spec waits for your yes. We gather the facts and escalate the same day.
AI in daily use
Every operator is trained to use AI daily.
Bots plus people
We run our own company this way: 25 bots, as of 20 October 2026. Each has one job, a list of things it may never do alone, and a person who says yes.
Bots do the work. People say yes.
Many AI agencies build the bots and leave. We run them, with people. A named person from our team approves what each bot prepares. What a bot may do alone is your choice, and it is written down. Which bots we set up depends on your business, so we agree that with you on a call.
Meet the teamSee the bots we set upRead our 12 rules for running bots
The method we can point to
This project was not for a technology company. It shows how we follow a specification and check quality.
Security
We work inside your stack. The data stays in your own systems, and we don’t download or store it.
A mutual NDA
Standard before we see anything sensitive.
A data processing agreement
We sign a UK GDPR data processing agreement before any agent gets a login.
Logins you control
We use individual named logins that you create and revoke. No shared accounts.
Who are your tech clients?
Our named proof is in food, property and e-commerce operations. For technology companies we bring the same method: people who follow your spec, check quality, and help you automate more over time.
So for technology companies we can speak about capability and method, not named clients. We can walk you through how we’d run it for you.
Questions people ask
We already have AI. Why do we need people?
AI is confident, and that is not the same as correct. Someone has to review the output, handle the edge cases and answer the customer who needs a real answer. AI alone isn’t magic; it only works when it’s built on real knowledge of your business, and that’s what our people bring.
Can you work behind our chatbot or support product?
Yes. We work behind AI chatbots, answering what the bot cannot and checking what it promised.
Can we try it first?
Yes. The first week is free: your person learns the work, shadows your team and drafts for your approval. Live work is charged from week 2, and you can stop with 30 days’ notice.
Who owns what you build for us?
You do. You own everything we create specifically for your business: the playbook, decision trees, reports and any client-specific automations. We keep our underlying tools and methods.
How much does it cost?
One all-inclusive hourly rate, typically 38 to 40% less than the true cost of an equivalent UK hire, and you only pay for the hours we cover. We send a written quote that shows the hours.
What do you not do?
We don’t fix your product or your code: we report bugs and patterns, and we don’t compete with your developers. We don’t make decisions that need your authority, and we don’t give legal or tax advice. We gather the facts and escalate.