Impact Heroes
Human in Command
You want to use AI confidently while keeping responsibility clear.
N.01Who we work withN.01
Who we work with
- Commissioners
- Boards, HR and L&D, training providers (including those in Poland’s Development Services Database), technology parks and clusters, universities and public bodies adopting AI.
- Participants
- Executives, team leaders, process owners, HR and L&D, and people working with AI tools and agents.
- Beneficiaries
- Teams that know what to delegate to AI; customers who encounter fewer errors; an organisation building AI literacy and human oversight, as described in Article 4 of the AI Act.
N.02The challengeN.02
The challenge
AI tools are already in teams. Agreements on who checks the output and who signs off decisions often are not.
The team leader is left unsure whether to slow down or move faster, and who will answer for a mistake. Responsibility for a decision should not disappear because a machine prepared it.
In Deloitte’s State of AI in the Enterprise 2026 survey (3,235 business and IT leaders across 24 countries), 20% of companies reported high readiness in people, compared with 43% in infrastructure.
N.03Why work with usN.03
Why work with us
We have worked with teams and leaders through change for years, including at Broadridge, Fundacja Greenpeace and Fundacja Edukacji Klimatycznej. We run a team of AI agents ourselves, so we know where roles break down in practice.
N.04What we doN.04
What we do
- Responsibility and rules for working with AIWorkshop: mapping AI use, autonomy levels, an exercise with a hidden AI error and a responsibility canvas for one process.pilot 2026
- Designing and managing a team of AI agentsWorkshop: agent roles and boundaries, backlog, safeguards, review and retrospective.pilot 2026
- Implementation reviewAn online meeting to review what you have put into practice.optional
We work on decisions and responsibility. This is not prompt training or a technical audit.
N.05How it worksN.05
How it works
- 01
Conversation and initial assessment
Where AI is already used and where agreements are missing.
- 02
Pilot in one process
The team agrees rules using its own work.
- 03
Programme for further teams
We take what works to other teams.
After each step, you decide whether we continue.
N.06What you take awayN.06
What you take away
- Rules for one of your processes: what AI does independently, where a person decides, who owns the outcome and when to stop.
- Five tools to reuse: an AI use map, autonomy scale, oversight matrix, responsibility canvas and team charter.
- One decision for Monday: what changes, who owns it and by when.
The cost of waiting
Without shared agreements, everyone makes up their own rules. The first serious AI error will reveal that no one owned it.
N.07Let’s talkN.07
Let’s talk about working together.
We can deliver a module in your programme, pilot with one team or co-design a programme for your business network.