For: Everyone who uses AI at work
After this, your team can check an AI answer before acting on it, keep company and customer data out of prompts, and use AI openly and within the law.
- Where AI answers go wrong, and how to trace one to its source
- What never goes into a prompt, and how to share access safely
- Saying when AI helped: disclosure that builds trust
- What the EU AI Act forbids or restricts, and who to ask
You leave with: A one-page checking routine, your own data rules, and an attendance record for your Article 4 file.
Helps you meet and document your EU AI Act Article 4 measures.
Plan this for your team For: Executives, heads of department, agile coaches
After this, your team can plan for a world where building is cheap and deciding and checking are the scarce part, with a one-page AI policy for their own area.
- The narrow part moved: from building to deciding and checking
- A one-page AI policy: access by need, who approves what an agent may do
- Licence, pay-per-use or in-house: getting more from the same AI budget
- Your first ninety days, written down
You leave with: A draft AI policy for your area and a budget comparison you can take to finance.
Plan this for your team For: Leaders, team leads, agile coaches
After this, your team can run a blameless review of an AI mistake and turn it into a check that stops it next time.
- Checks that lie: when a passing check is not a working one
- A blameless review, step by step, on a real case of yours
- From finding to guardrail: the check that blocks it next time
- Who owns the check, and how often it is re-tested
You leave with: One reviewed incident from your own work and the check that now guards it.
Plan this for your team For: Business users, leaders, analysts
After this, your team can stage a structured debate between AI-simulated viewpoints on a real decision, and keep the human judgement where it belongs.
- Why one AI answer is the weakest kind
- Casting the room: personas built on published thinking, clearly labelled as simulated
- Running the debate: divergence, friction, convergence
- Deciding, and writing down why
You leave with: A decision of your own, argued from four sides, with the reasons kept.
Plan this for your team Advanced · 2 days · up to 8 · PL or EN
Find your product’s problems before customers do
AI walkers, then real customerssimulated
For: Product managers, UX, product owners
After this, your team can send AI walkers through a test version of their product, keep only findings a second check reproduces, and take those hypotheses to real customers.
- Personas from your own research, and their limits
- Walking the product: findings tagged heard, plausible or walker-only
- Adversarial checks: re-derive every claim before it ranks
- From hypotheses to customer interviews, and a decisions page where every pick has a "not now"
You leave with: A ranked list of hypotheses about your own product and an interview plan to test them.
Two days, breaks included.
Plan this for your team For: Developers, IT and platform teams
After this, your team can put agent work behind gates, catch the checks that lie, and route each task to the cheapest model that does it well.
- A fresh agent per task, with a written brief, and done means the gate passes
- Checks that lie, and testing the way it will run
- Memory and handoffs: what an agent keeps, and what it drops
- Secrets, costs and model routing: frontier, everyday or your own small model
You leave with: A working gate on one of your own repositories and a routing rule for your models.
Two days, breaks included.
Plan this for your team Price
From 9,000 PLN net per day, for a group of up to 12.
- Two-day trainings from 16,000 PLN net.
- AI literacy for larger groups from 7,000 PLN net per day, for up to 20.
- Remote: 20% less.
Custom-tailored trainings. Need something else: a mix of these, a programme over several months, or a training built around one process of yours? Tell me what should change and I'll propose a programme and a price.