Who is accountable when the AI gets it wrong?

Companies believe they are adopting AI. In fact, they are quietly shifting responsibility onto their developers. I turn that shift into a decision, so you can scale AI without leaving responsibility, ROI and risk ungoverned.

Boardroom to code.

For organisations whose software already contains AI – before the next audit is due, an incident occurs or the next EU AI Act milestone arrives.

Book an introductory call
71%
say accountability for AI-caused defects is unregulated and lands on the developer
40%
name unclear governance as the No. 1 bottleneck, ahead of technology
6%
feel relieved at day's end; 67% experience denser, more demanding days

Source: Python Software Association Germany · a survey of 383 software professionals · June 2026

Most companies do not have an AI-governance problem.They have an unresolved architecture of accountability.

Unresolved means: responsibility for AI shifts in four forms at once. Legal liability, internal accountability, operational responsibility, sign-off.

The three-stage model

From the diagnostic to the operating model

Diagnostic → Decision → Operating model

Boardroom-to-Code means
  • Board: strategy, risk, ROI
  • Legal and compliance: liability, EU AI Act
  • Architecture: ownership, review, evaluation
  • Code: engineering's daily reality
01

The diagnostic

Stage 1: the risks are prioritised

I read your organisation across every level, from the boardroom to the code, along the framework. It shows where responsibility is unclear and where risk concentrates.

02

The Boardroom-to-Code Session

Stage 2: the board can decide

A focused session with your leadership team. Priorities, responsibilities and trade-offs get decided. It ends in the Executive Briefing.

03

The operating model

Stage 3: ownership becomes dependable

The design of the operating model: ownership and sign-off, review and eval standard, connected to your legal and compliance function, EU AI Act included. Documented board-ready.

Governance is not a brake; it is the precondition for scaling.

Stage 1 · The framework

Five dimensions where unassigned responsibility surfaces.

  • AccountabilityWho signs off AI code, and does that hold up in an audit?
  • ROI realityDoes the speed still pay after review, rework and risk?
  • Value migrationWhere does the value of your engineering move, and does the role model follow?
  • DeskillingIn three years, will your organisation still understand its own code?
  • Shadow AIWho decides on tools that are already in use?

The productivity is real. Quality assurance, responsibility and load stay unmanaged: a leadership question, not a tooling one.

Stage 2

The Boardroom-to-Code Session

The decision. I bring the diagnostic into a focused session with your leadership team. This is where the decisions get made that no one has made so far.

How it runs

  1. The findings from the diagnostic, prioritised along the framework
  2. Responsibilities and trade-offs, decided with the room
  3. Executive Briefing: which body decides what, responsibility options, review tasks

Who decides

The people it concerns: the board or CIO and CTO, engineering leadership, compliance, security.

The basis

The diagnostic from stage 1: your situation, read across every level.

Why me

I translate between the board and engineering until a decision stands.

Proprietary evidence, both languages

I co-authored the survey by the Python Software Association Germany: solid data on what developers actually experience with AI. I translate it into board language (risk, ROI, compliance, workforce) and back into architecture and code. The craft of moderating decisions between the business and technology is twenty years proven; only the subject is new.

Independent by design

No software, no licences, no commissions. My recommendation follows your situation, not my revenue.

Three mandates, three outcomes:

Asset management: release cycles from three months to three weeks → Public infrastructure: workforce planning from annual to continuous → Research & development: 10,000 documents, search in seconds, on-premise →

The right call in week two saves months of misdirected development, and a risk no one chose to carry.

Whitepaper

Read first, talk after: the decision paper.

The full analysis of the survey as a PDF: four findings, three trade-offs, four tests for your next leadership meeting. 16 pages, straight to your inbox.

Request the decision paper

Your data is never shared with third parties.

Next step

Which gap is costing your organisation the most right now?

The shortest path is the introductory call, for us both to see whether the session fits your situation. I reply personally either way.

Your enquiry