Leaders shape the conditions in which AI creates value
Training helps employees explore tools and assess their outputs. It does not, on its own, resolve unclear priorities, slow approvals, conflicting measures or inaccessible data.
AI implementation also involves leadership decisions: which business challenge to address, who can change the workflow and who owns the end-to-end outcome. Below, we distinguish research findings, Microsoft’s proposed action sequence and LeanAI’s practical interpretation.
What does the research show, and what does it not establish?
67% and 32%. In Microsoft’s analysis, organisational factors such as culture, manager support and talent practices had greater relative importance in relation to respondents’ reported AI impact than individual factors. These are statistical model importance measures, not shares of profit or financial return, and they do not establish causation.
13% of surveyed AI users say they are rewarded for redesigning work with AI even when intended results are not achieved. For leaders, this raises a practical question: do teams have time and clear boundaries to test a different way of working while retaining accountability for outcomes?
The survey covered 20,000 knowledge workers who use AI at work across ten countries. It does not represent all workers or Lithuanian companies. The analysis relies on respondents’ own assessments and should not be equated with measured company financial impact.
Choose one workflow that matters to the business
The guide recommends starting with work that is important, understood and sufficiently bounded for a credible test. For example, improving order approval may matter more than automating summaries across the business if approval delays are the main constraint on delivery.
Record a baseline before the trial: volume, cycle time, errors, rework and cost. Assign one business leader who can resolve cross-functional dependencies. Agree in advance what evidence would justify scaling, redesigning or stopping.
Agree on human and AI responsibilities
Define three parts of the workflow: what people do themselves, what they do with AI and what a system may execute independently. Agree on data access, quality checks, exception handling and decisions requiring human approval.
Data preparation, employee learning and ongoing oversight belong inside the trial. Include these costs in the assessment and connect released capacity to a specific use.
A possible first 90 days
Days 0–30: choose a process, understand its workflow and establish a baseline. Check data availability, sensitivity and the agreements needed to change the work.
Days 30–60: redesign the workflow and run the smallest useful test. Set an accountable owner, human and AI decision boundaries, and an end date.
Days 60–90: review the evidence and decide whether to scale, redesign or stop. Establish oversight and spending controls before expanding. This is a planning example from Microsoft’s guide, not a promise of financial returns within 90 days.
Three situations in which to look for a first trial
Do you discover a significant order delay too late? Does the same blocker repeatedly surface in meetings across departments? Does an investment proposal depend on assumptions that nobody has fully checked?
We have developed three illustrative LeanAI scenarios around these situations. They cover the AI task, the process change, decision boundaries and how to assess value.
Read three AI application scenarios for manufacturing leaders →
How this connects to LeanAI’s approach
LeanAI starts with a business challenge that matters to leadership. Lean helps identify the constraint on results and focus the team on process change; AI expands the range of potential solutions. The IMPACT method connects analysis, trials and embedding an effective way of working.
The first question is which process improvement would make the greatest difference to your organisation now. Microsoft’s material adds context to that discussion; a practical action plan needs to reflect your circumstances.
Understanding the sources
The research figures come from the 2026 Work Trend Index and its methodology. The Frontier Leader is a Microsoft executive guide combining research insights, recommendations and Copilot examples. Pages 5–6 and 23–26 are particularly relevant here. Technical suitability and integration availability need separate assessment.
Sources
Microsoft Work Trend Index 2026 — report and methodology ↗Microsoft The Frontier Leader — publication registration page ↗
Which challenge most limits your business results?
The first step is a meeting with the leader to understand the current situation and expectations.
Let’s discuss your business challenges →AI and Lean implementation and practical corporate training →IMPACT method →