Platform & Team

Forward Deployed Engineers

One of our senior AI engineers works inside your team and is accountable for a result you agreed together, not for a list of tickets. They start with your real data, your real process and your real constraints, and they finish when the thing is running and your people are using it.

What the engineer can build for you

One person, senior, with the range to take a job from a first conversation to something running in production. What they build depends on what the work needs.

01

Assistants on your own documents

Answers pulled from your own contracts, manuals and records, with the source attached so anyone can check them. Nobody has to trust a black box.

02

Models for your hard cases

When a general model keeps getting your cases wrong, we build one that does not. Trained on your history, measured against the decisions your people already make.

03

From prototype to production

The unglamorous half: deployment, access rules, logging, monitoring and cost control. A prototype that nobody can run is not a result.

04

Design that fits what you already run

We work with the ERP, the CRM and the spreadsheets you already pay for. Replacing your systems is almost never the cheapest way to fix a process.

05

Numbers you can act on

Your data cleaned, joined and put somewhere everyone reads from, and then forecasts built on top of it that hold up when a real month goes wrong.

06

Deciding what comes next

We watch what people actually use, not what the plan said they would, and tell you which piece of work is worth funding next.

What you end up owning

Everything the engineer produces is yours, in your repository and your accounts, from the first day.

A working system in production

Running on infrastructure you own, used by the people whose job it was meant to change.

The one page we agreed in week 1

The outcome, the measure and the failure conditions, updated with what we actually learned.

An honest measurement

What the process cost in hours before, what it costs now, and where the system still needs a person.

Documentation your team can use

How it works, what it costs to run each month, what breaks it and who to call.

A short list of what is worth doing next

Based on what we saw inside your business, with the ideas we think are not worth funding named as well.

Platform & Team

How the first two months go

01

Fit call

Thirty minutes with the engineer who would do the work. If we do not think we are the right answer, we say so on that call.

02

Week 1: understand

We watch one full cycle of the work, get access to the systems it touches, and agree one page: the outcome, how it is measured, and what would make it fail.

03

Week 2: something running

A first working version on your own data, in front of the people who will use it. It will be narrow and it will get things wrong, and those errors write the rest of the plan.

04

Month 2: production and handover

Wired into your systems, with access rules, logging and a way for a person to check or override it. Written down so your team can run it without us.

How the engagement is shaped

A monthly fee for one engineer, not an hourly rate and not a price per ticket. The number belongs in a conversation about your actual situation, so let us have that conversation first.

30 minutes · free

Fit call

A straight conversation with the engineer who would be embedded, about the job you want done and whether this is the right way to do it.

  • You talk to the engineer, not to a salesperson
  • We name the cheaper option when there is one
  • You leave with a sharper description of the problem
  • No document to sign and nothing to pay
3 months · monthly fee

First result

The minimum that gets to something real. Understand in week 1, something running in week 2, in production and handed over by month 2.

  • One senior AI engineer inside your team
  • One agreed outcome, measured the way you agreed
  • Everything in your repository from the first commit
  • Nothing to renew if the result is enough
Monthly · 30 days notice

Ongoing

For companies that find a second and a third job worth doing once the first one lands. The same engineer, the same accountability, month by month.

  • Continues after the first three months
  • Thirty days notice on either side, no exit fee
  • Scale to a second engineer when the work needs it
  • Stop when your own team can carry it

Who this is for

You know the job, not the how

You can point at the process that costs you the most hours. You do not have anyone in house who could build the thing that fixes it.

You have a strategy and nothing running

Someone delivered a document about your AI opportunities. Nobody has since put anything in front of the people who do the work.

Hiring is too slow for this

You would need months to find a senior AI engineer, and you are not sure yet whether the work justifies a permanent hire.

The exceptions are the hard part

Your process has rules that nobody has written down and cases that never look the same twice. Fixed scope written up front would be a guess.

What we build with
Python
TypeScript
PostgreSQL
Anthropic & OpenAI APIs
Open models you can self-host
Docker
AWS, Azure or your own servers
The ERP and CRM you already run

Questions, answered

How is this different from your Team Extension service?

Team Extension gives you capacity: engineers who work the backlog your team writes. A forward deployed engineer is accountable for a result instead. They decide what to build with your named decision maker, based on what they see when they watch the work being done. If you already know exactly what needs building, Team Extension is cheaper and simpler, and we will tell you that.

Why not a fixed price project?

Because with this kind of work the scope is genuinely unknown until you run something against last month's real cases. A fixed price either gets padded to cover that, which makes it expensive, or gets held to the letter, which makes it useless. Our Requirement Analysis and MVP Deployment engagements stay fixed price, because there the answer really is knowable up front.

What do we have to provide?

One named decision maker who can say yes and is available for a short call each week. Access to the systems and data the work touches, arranged in week one. A couple of hours with the people who do the job today. And honesty about the rules you have to follow, including the ones nobody has written down.

What happens to the work if we stop?

It stays with you. Everything is in your repository and your cloud accounts from the first commit, and the handover documentation is a deliverable, not a favour. After the first three months there is thirty days notice on either side and no exit fee.

What will you not promise us?

We will not promise an accuracy figure before we have seen your data. We will not promise a fixed price for an outcome nobody can describe yet. We will not promise that AI is the answer, and if a rule, a script or a process change is cheaper we will say so. We are not an on call team, and we do not make your decisions for you.

Want to know if this fits?

Thirty minutes with the engineer who would be embedded. You describe the job you want done, and you get a straight answer on whether this is the right way to do it, or whether something cheaper would work.

Book a call with Krzysztof