Independent AI advisory · Leadership and engineering

First CTO of a company later valued at $2.3B.
Now I build the system your AI runs on.

Your organisation adopted AI fast: a tool here, an automation there, agents wired into different systems. Now you have dozens of experiments and no system. I work with the leadership that decides and the engineers who build.

A small number of engagements each quarter. Write to me

Portrait of Maciej Lotkowski

Track record

  • $2.3B Valuation reached by Jobandtalent, where I was first CTO
  • 2M+ People on Jobartis, the job platform I co-founded
  • SoftBank
    Kinnevik
    BlackRock     Backers of Jobandtalent

Your AI tools work. Your AI system doesn't.

About

Dozens of tools, no shared data layer, no governance, no way to tell what's working. The experiments multiplied. The results didn't.

The fix is not another tool. It is one system underneath all of them: every agent and automation working on the same governed company data.

From CTO to advisor

I was the first CTO of Jobandtalent. I built the engineering team, the architecture and the data infrastructure the company scaled on. It later raised $500M from Kinnevik and SoftBank at a $2.3B valuation, with debt financing from BlackRock.

Then I co-founded Jobartis, a job platform used by more than 2 million people across sub-Saharan Africa.

Today I advise on AI from the same position: which projects will pay off, which will quietly stall, and what has to be true of the data underneath.

Services

Two engagements. One system.

One for where the decisions get made. One for where the software gets built.

Most engagements start with a day on site. What follows depends on what the day finds.

For leadership

AI Strategy for Leadership

The problem

Initiatives overlap, tools multiply, and no one at the top can say which of them earn their cost.

I work directly with your C-suite, founders and VPs. You leave with four things:

  1. Readiness assessment

    Where you stand, against your own business goals.

  2. Opportunity map

    Prioritised, with expected ROI on each.

  3. Leadership alignment

    One shared view of AI vision and governance.

  4. 90-day action plan

    Next steps, each with a named owner.

Start with leadership

For engineering

Autonomous Delivery

The problem

Your developers got faster. Your organisation didn't.

A staged programme, not a rebuild. Four stages, from how your team works today to agents running routine delivery. Your team owns the rules at each one.

  1. Literacy

    A shared baseline. Everyone working the same way.

  2. Method

    Spec-to-code as a repeatable process, not a personal habit.

  3. Gates

    Automated checks that let work move without a human in every loop.

  4. Autonomy

    Agents that run the routine work end to end, under rules your team owns.

Building software is communication — and AI is now part of the conversation.

Start with delivery

Clients

In their words

  • Worth it. The ROI? Speed, and a team that's confident in the work.

    Artur Czemiel · CEO, Aexol

    Likelihood to recommend: 10/10 · Impact on team effectiveness: significant

    Client's ratings at the close of the engagement

Speaking

On the record

Talks for engineers, executives and auditors.

Ventures

What I'm building

Two products, built alongside the advisory work.

  • Agent Lake

    An AI operating system: a governed foundation for running AI agents on company data, with control over access, provenance and security.

    I build Agent Lake because I kept recommending an architecture nobody was shipping. Advisory clients are under no obligation to use it. The pattern matters more than the product.

    agentlake.dev 

  • AI for Legal Teams [stealth]

    Document workflows for law firms that have outgrown Word and OneDrive.

Contact

I take on a small number of engagements each quarter.

If you have the experiments but not the system, write to me directly.