Implementing AI without losing yourself 5/5
We show that an AI compass on one page clarifies purpose, values, and responsibility. By building it together, we align governance and culture before scaling.
Towards an AI compass on one A4
Build a compass, not a manual.
Here we bring everything together on one A4: choices that guide tomorrow and shape culture today.
We brought together three management themes: (1) purpose, values, and boundaries, (2) AI as a socio-technical system, and (3) governance, risks, and learning architecture. It is time to capture this in an AI compass on one A4 — not a technical blueprint, not a legally exhaustive document, not a marketing story, but a clear set of choices.
Four core questions for your AI compass
What may AI be used for?
Which values and boundaries are non-negotiable?
How do we distribute roles, power, and responsibility (including overruling)?
How do we learn from what AI sets in motion (rhythm, incidents, adjustment)?
Suggested structure (4 × half a page)
Our purpose with AI – 3–5 sentences about its contribution to purpose and strategy; 2–3 sentences about what AI will never replace.
Values and red lines – 3–5 values translated into principles (“always human final judgment,” “right to explanation”), plus 3–5 no-go’s.
Roles and relationships – ultimate accountability, protected dissent, handling differences between model and professional judgment.
Governance and learning – executive owner, use-case classification (learning zone/critical zone), reflection rhythm, and escalation paths.
Why this compass itself is culture development
The process is the intervention. You talk about identity (who do we want to be in an AI-rich world?), make implicit assumptions explicit (trust, control, professionalism, mistakes), and show that management leads in reflection, not only in ambition. In that way, AI becomes a driver of maturity rather than an accelerator of alienation.
What now – concrete steps
Small core group, short timeframe – 2–3 MT members, HR, IT, and someone from operations; two sessions of two hours for version 1.
Mirror from the organization – present the draft to professionals, the works council, and if possible an external critical friend; ask: what becomes easier, what becomes more tense?
Formalize it — and keep questioning it – adopt it formally in MT/board; immediately schedule an evaluation in 6–12 months: what happened, what did it set in motion, what do we refine?
Reflective closing
When you look back in five years at your AI choices, do you want to say we were early and fast — or we remained true to who we wanted to be, even when technology moved faster than we did?
What one step will you take this month to make that second sentence true?