Three doors

Every implementation starts with one bounded scope and ends with acceptance. Pick the door that matches what you want to improve.

Automating one process

I take one repetitive piece of work off your hands and leave a working, tested workflow with acceptance criteria — not an endless transformation plan. Supplier data, documents and e-mail, invoice flow, sales enquiries.

Start with a pilot on your own data →

AI and MCP integrations for products and teams

A chosen AI assistant can safely use the functions of your system — with permissions that can be traced and behaviour that can be tested. For teams that own an API and the deployment decision.

AI agents in your development team

I turn the use of coding agents into a repeatable, reviewed engineering process in your repository — measured on the team’s real work, not on the number of lines generated.

Regulation and where to start

AI Act technical compliance sprint →
Your lawyer sets the scope, I implement it: a register of systems, content marking, documentation and the Article 4 workshop.
AI readiness and opportunity audit →
An optional first step: which processes are worth improving, what data they need and when plain automation is enough.

See all services →

Not sure where to start? See what AI implementation looks like step by step — from checking the process to handing over to your team.

Where this comes from

What can be checked — and one experience I put my own name to.

20+ years

on production systems — the founder's experience

My name is Krzysztof Ploch. For over 20 years I have designed and built software: production systems for companies from a few dozen people to corporations with more than 20,000 staff, integrations between systems and process automation. My AI implementations rest on that experience, not on courses about AI.

One caveat about numbers: at the last company I worked for, we introduced AI agents as the core way of working, and within a year more than 70% of the code was produced with their involvement. That is my experience from one team, not a benchmark — and I do not promise it in your company. Every process is measured separately, on your data, before anything reaches production.

2024

The year the business was registered — details on the contact page

1 process

What I take on to start: one scope, acceptance criteria, handover

PL and EU

Run from Lublin and Warsaw, working remotely across the European Union

How I work

Four stages — with a point where you can walk away before paying for a full implementation.

Viability check

A conversation about one process: what repeats, what it costs and who knows it. I also check whether a tool you already have solves it more cheaply. If it does not add up — I say so.

Scope and acceptance criteria

We agree what the pilot covers, on which data, who owns the process and how we will know it works. The scope and the criteria are written down before anything is built.

Integration and testing

I build the solution inside your systems, test it on samples and run it in a limited scope — with a person approving the results.

Handover

You keep a working system, documentation and tests, and the team practises with it on real tasks. I do not run standalone training — the team's self-sufficiency is part of acceptance.

Why ploch.ai?

You work with the person who designs and builds the solution — from the first conversation, through the process audit, to going live and handing over. There is no account manager passing the work on. For scopes that call for it, I bring in collaborators with the skills that scope needs.

Automating a process in an operating business is, for the most part, an integration and maintenance problem rather than a model problem. So I start from your systems, your data and the people who know the process — not from choosing a tool.

I am neither an agency nor a software house. An agency sells a campaign, a software house sells hours. Here the starting point is a specific process and the question whether automating it adds up. Some conversations end with a recommendation not to run the project.

One process to start, with acceptance criteria — not a transformation plan
Proven on production systems, not in slide decks
Your team learns to own it, not to depend on me

Who is behind ploch.ai, since when and under which registration details — on the about page. What I don't do and when I advise against a project — on the refusal list.

What stays in your company after the implementation

A working system in your environment, with a repository or a configuration export — no dependency on my access.
Operating instructions, known limitations and a rollback procedure, written for the person who will run the system.
Regression tests and acceptance criteria, so you can check that a model or API change has not broken anything.
A team that has practised on real tasks, not on a course — and a named person in the company who takes the topic further.

Ethics

Bringing AI into the work does not have to mean redundancies. I run implementations so that the company goes through the change without cutting jobs, or at the lowest possible cost — for ethical reasons I do not want this industry to be associated with job cuts.

For people whose work will increasingly be done by AI, I plan alternative tasks together with the company — including those related to maintaining and developing the solutions that have been put in place.

Your employees know the company and its processes. AI does not know the nuances — they do. In my experience it is better to invest in the company's throughput and pace than in reducing headcount.

Write to us through the form

The message goes straight to our mailbox. If you would rather e-mail or call, both are above.

Information on data processing

The controller of your data is Actum Krzysztof Ploch, ul. Irzykowskiego 3/15, 01-317 Warsaw, Poland, contact@ploch.ai. We process the data from the form to reply to your message and — if you ask for one — to prepare an offer (Art. 6(1)(b) GDPR: steps prior to a contract, at your request), and in our legitimate interest in handling correspondence and protecting the form from spam (Art. 6(1)(f) GDPR).

The message goes to our mailbox and is not stored in any database on the site's server; the recipients of the data are the hosting provider and the e-mail provider. We record a hash of the IP address on every attempt to send, and — if you have JavaScript enabled — already from the moment you start filling in the form; we keep it for one hour after the last use, solely to limit the number of submissions; the hash does not reveal the address (the key needed to compute it never leaves the server), and once the hour has passed it is deleted at one of the next uses of the form by anyone. We keep correspondence for as long as needed to handle the matter and thereafter until any related claims become time-barred.

You have the right to access, rectify, erase and port your data, to restrict its processing, to object to processing based on legitimate interest, and to lodge a complaint with the President of the Personal Data Protection Office (UODO). Providing data is voluntary, but without an e-mail address we cannot reply. Details are in the privacy policy.

I put AI to work where it has a specific job to do.