You need intelligence to use intelligence.

No-bullshit AI strategy and sparring for executives and product leaders. Twenty years in AI is enough to see hype for what it is – and to think most companies still underestimate what comes next.

Book a workshopInvite a keynote
Jan Romportl, black and white portrait against a concrete wall
  • Czech AI Personality of the Year 2024
  • AI Personality 2025 – Security
  • Ph.D. in AI
  • former Chief Data Scientist & Director of AI Centre, O2 / PPF (founder of Dataclair.ai)
  • built & exiting Elin.ai
  • Chairman of the Advisory Board, Czech AI Association
  • Prague AI Safety Meetup organizer since 2018

What I actually think

AI will not save your company. Clear thinking about it might.

AI will take more jobs than it creates.

I cannot tell you the timing, and neither can anyone quoting you a percentage. What I can do is go role by role with you: which tasks a model already does at a quality your customers accept, which ones a person has to sign off on for legal or reputational reasons, and which ones people want done by a person because a person did them. Care, teaching, negotiation and work that carries personal responsibility go last. Plan your hiring and training around that.

A copilot for everyone is not a strategy.

Buying a licence for everybody hands the question to each individual employee, and most of them have no way to answer it. The work that makes AI usable sits in IT: connecting models to the systems where your data actually lives, deciding who may deploy what, and checking whether a use case does what it claimed before it reaches customers. That is what I help companies put in place.

Most AI takes are hype or dismissal. I sell neither.

I have spent years calling out inflated claims, and I still expect AI to change your industry more within ten years than your current plan assumes. Both of those hold at once, which makes the useful question one of sequence: which parts of your business AI reaches first, and how soon. That is what I work on with clients.

What I do

Four ways to work with me.

Executive workshop

Half-day

CEO, C-level, board

Decide which AI bets you are making this year and which you are dropping.

You leave with a use-case shortlist, priorities and a 30/60/90-day plan.

Ask about this →

Team workshop

1–2 days

Product leaders and their teams

Unblock hard problems, find concrete opportunities.

Decide what to build and what to kill – with owners and next steps attached.

Ask about this →

Keynote

60 min + Q&A

Leadership conferences and partner meetings

High-signal sense-making about where AI actually is.

One hour on what it means for your industry, then as long as the questions keep coming.

Ask about this →

Advisory sparring

Retainer

Prepaid office hours after a workshop

Someone outside your reporting line to test a decision against before you commit.

You bring the decision you are about to make and we go through what would have to be true for it to work, and what would break it. I advise and challenge. I do not run your programs.

Ask about this →

Selected clients

Škoda Auto · Mastercard · Allwyn · PPF · O2 · T-Mobile · Deloitte · PwC · Home Credit · CETIN · PPF Banka · Czech Banking Association · Hewlett Packard Enterprise · TietoEvry · Bel Brands · Livesport · Alza · Prague Airport · Emma Capital · Investika · Seyfor · NAKIT · Cloudfield · Geetoo · Rixo · Railway Research Institute

Current frontiers

Agentic safety & AI security

Autonomous agents are entering companies faster than anyone can govern them, and AI safety engineering is converging with cybersecurity. This is what I am working on most right now, together with cybersecurity teams: what an agent may do on its own, who authorises it, and what gets logged.

AI, minds and mental health

I built an AI startup protecting young people's minds on social media (Elin.ai), led a research group at the National Institute of Mental Health, and teach digital resilience. If your AI questions are about wellbeing, children, attention or trust, this is the area I have worked in longest.

The enterprise AI backbone

Model access, integration, governance, evaluation – the unglamorous layer that decides whether AI creates value or theater in your company. See stance 02 above; this is the consulting core.

Why me

Where this comes from.

Academia, then Chief Data Scientist and Director of the AI Centre at O2 / PPF, then building Elin.ai. I have run AI research, killed AI projects with a budget attached, and shipped an AI product to real users.

High-agency workshop style: intense, creative, concrete. You get real thinking in the room and real outputs on the table – shortlists, priorities, decisions – not slideware theater.

I treat safety and governance as part of the design. If you can show how an AI decision was made and who is accountable for it, you can put AI into the processes where the money and the risk actually sit.

Jan Romportl, relaxed black and white portrait

Not for everyone

Let me save us both some time.

What I do is not an AI-tools productivity training, and I am not a project manager for hire. I will not run your rollout, own your execution or manage your people. I come in when a decision is stuck and leave once it is made.

The best fit: smart, high-agency teams who want real thinking and decisive next steps. If that is you, we will get along well.

Ready when you are.

[email protected]