01 · Implementation problem
Why AI projects often fail
Seven pilots, local champions, first results – but the expected lever never materialises. This pattern repeats across SMEs. In practice, three causes appear.
Cause 1
Tools without processes
AI is introduced as a tool – but the underlying processes stay unchanged. Copilot in engineering, but the review process for drawings is just as slow as before.
Cause 2
Data without use
Data is available – but decision paths do not use it. Reports are generated, yet leaders still decide based on gut feel.
Cause 3
Technology without leadership
AI systems are introduced, but leaders are not trained to interpret, challenge and own AI outputs.
Observation
None of these obstacles is a technology problem. They are organizational and leadership tasks. AI does not fail because of the tool – it fails because no one decided how the tool belongs in everyday work.
02 · Three levels
What effective AI impact looks like
AI only unfolds its impact when three levels work together. Technology alone changes no organization – but the right combination of structure, processes and people does.
| Level | What this means in practice |
|---|---|
| Organization | Structures and decision paths that enable AI-supported work. Clear roles: who may decide what with AI? Who owns AI outputs? |
| Processes | Workflows in which AI creates genuine value – not just complexity. AI is integrated into existing flows, not bolted on as a foreign body. |
| Leadership & skills | Capabilities and mindset to use AI responsibly. Leaders understand outputs, challenge them, communicate limits. |
AI as a tool project
- Roll-out by IT – communication afterwards
- Departments experiment individually
- Success = pilots launched
- Governance as an afterthought
- One-off, central training
AI as an organizational project
- Strategic MD commitment from the start
- Prioritisation by business impact
- Success = measurable KPI improvement
- Compliance and governance built in from day one
- Capability building in the context of real work
03 · Path to Resilience
Six steps to AI impact
The Path to Resilience follows a clear logic: first understand, then focus, then implement, then learn. Six steps make the transition plannable.
- 01
Analysis and diagnosis
Problems and patterns become visible – data-based, not anecdotal. Where is AI already running? With what impact? Where are the biggest gaps?
- 02
Strategic focus
Many options become a few levers: which use cases serve strategy, value creation and resilience?
- 03
Use-case prioritisation
Assessment by impact, feasibility, strategic relevance and regulatory risk. Result: a prioritised map.
- 04
Pilots with impact
Pilots are set up so that impact becomes measurable – with clear KPIs, owners and learning loops.
- 05
Scaling into the organization
What works is transferred into structures, processes and leadership. AI becomes part of the operating model, not an attachment.
- 06
Early-warning system
Changes are spotted early and steered actively – regulatory, technological, market-side. The Early Warning Canvas keeps you informed.
04 · Diagnostics
Four questions that start the impact
The most common mistake in AI projects: starting with the solution instead of the problem. Anyone introducing AI without knowing where the biggest levers are invests in activity instead of impact.
Where do we lose time?
In which routines do we lose the most time today – and which of them could AI sensibly take over?
Which decisions are too slow?
Which decisions are made too slowly with us – and why? Where do data gaps or unclear ownership block them?
Which data do we not use?
Which data do we already have – and is it actually used to improve decisions?
Where does AI already work?
Where is AI already active with us – and which of these applications produce demonstrable impact rather than mere activity?
| Assessment dimension | Guiding question |
|---|---|
| Impact | How big is the lever for efficiency, quality or speed? |
| Feasibility | How complex is the implementation – technically, organizationally, in terms of data? |
| Strategic relevance | Does this use case serve the core strategy of the company? |
| Regulatory risk | Which compliance requirements does this use case carry? |
05 · Hybrid organization
Humans and AI in interplay
The goal is not full automation – it is a hybrid organization in which human judgement and AI capabilities are deliberately combined to deliver better outcomes than either could alone.
Always human
- Ethical judgement in edge cases
- Crisis decisions under time pressure
- Building trust with customers & partners
- Leadership in change processes
- Final accountability for outcomes
Ideal for AI
- Repetitive analysis & reporting
- Monitoring and proactive alerts
- First capture & preparation of information
- Consistent execution of defined processes
- Scaling without capacity limits
AI as an amplifier, not a replacement
AI is not an add-on project. Properly embedded it supports every phase of the process – from pattern recognition in diagnosis to monitoring in the early-warning system.
06 · Case study
Consulting: from disorientation to an AI strategy
A mid-sized management consultancy was under heavy change pressure from AI – clients asked for it, competitors were already using it. Internally, disorientation prevailed: everyone talked about AI, nobody decided.
to an anchored AI strategy
From Resilience Check to prioritisation to operational embedding – with clear MD commitment.
faster proposal creation
In sales through AI-supported research, structuring and text modules – at consistent quality.
time for internal research
Through AI assistants with access to structured internal knowledge instead of manual search.