We move your company's routine work onto AI agents, and keep them working.

Two founders from fintech and healthtech who run their own work on agents.

You pay for the result: if your first agent misses the target in the contract, your upfront payment comes back.

Book a free call

14 supplier emails sorted
3 replies drafted for you
1 invoice doesn't match its order

07:10

What an agent can take off your plate

"I spend my evenings in email."

It sorts client and supplier mail every morning and drafts the replies.

"Who still owes us?"

It matches invoices to orders and payments, and lists what is unpaid.

"Tell me in one message what happened overnight."

A morning brief in Telegram: money, orders, and what needs you today.

"The weekly report eats my Monday."

It builds the report from your spreadsheets and CRM on schedule.

"I meant to write back."

Leads, quotes and suppliers get their follow-up on time.

"We copy it from one system to another by hand."

It moves the data between your systems, in a browser if it has to.

We automated our own companies first

Andy Avramenko

Fintech

Founder of Ohmyfin, ex-Doctis, ex-IBM

Slava Sazhin

Healthtech

Founder of Doctor at Work

Both run their own work on agents every day. You talk to us, not to a support queue.

More about us

How we start

1
Tell us about the work

A short request, and you pick the time in our calendars yourself.
2
A free 30-minute call

We look at your week with you and pick the first task for an agent.
3
Your first agent, paid for the result

€200 upfront, €400 once 80% of that person's routine runs on AI. Otherwise the €200 comes back.
How paying for the result works
4
We keep it running

From $57 a month for each agent we look after, plus $14 for its server.
5
A team, or more work

Priced after the first agent works, per person or per process.
Book a call

Your Claude plan is your own, paid to Anthropic, not to us: budget 100-200 euro a month.

Fair questions, straight answers

Why not ChatGPT's or Claude's own agents?

Ours runs on a machine of its own, with your files, on your own Claude account, with a mailbox and a Telegram bot of its own, and two people who build it with you and check it.

What if it does something wrong?

For each task we agree what it does alone and what waits for your OK. Only senders you list give it orders, a run cap stops a loop, and you can cut its access at any time.

Does my data leave the EU?

The server and its files are in an EU datacentre. What the agent sends to the model goes to Anthropic, a US company, under your own account and its terms.

What if it doesn't work for my business?

We say so on the call. If we take it on and miss the target through no fault of yours, the first payment comes back.

Is there a contract and an invoice?

Yes, from our UK company. EU businesses with a VAT number pay no VAT. Our data processing agreement is part of the Terms.

Rather build agents yourself?

Rent a server with Claude Code and run your own.

Servers

Which task should we automate first?