AI Customer case: The leadership imperative
- Tom Hansen
- Aug 31
- 4 min read
This customer case is for the more advanced user. We recommend that you start listening to podcast, and then read the article.
Leaders face a simple pressure that never feels simple. The next role arrives faster than the calendar allows, while the work in front of you does not pause. In one case, a senior leader had four months to step into a much larger job. Postponing preparation would weaken the start. Generic training would consume time without building the specific judgment the role required. The problem to solve was direct. How do you turn advanced AI into a daily amplifier of an experienced leader’s method, inside the real work, with proof that the preparation improves week by week.
The third path
The answer grew from a practical constraint. Preparation had to live inside the day rather than beside it. We built a microchange engine that sits in a dedicated project space and proposes one small action every morning at eight Copenhagen time. Each action takes five to ten minutes and is completed where the context already lives. The work is captured as it happens. A Socratic tutor in the same space asks for reasoning, decisions and next moves.
Fortnightly coaching draws on this record rather than on recollection. Progress is visible because it is grounded in artefacts.
Evidence first, then narrative
The project treats claims as hypotheses that must be checked against sources. The model is instructed to show where facts come from, to log conflicts when files disagree, and to mark uncertainty that requires a follow up question. This forensic posture is not a stylistic preference. It is the only reliable way to connect learning with execution when time is short.
Leaders do not need an avalanche of output. They need a line of reasoning they can trust, with traceable origin and a clear next step.
Selecting the right prompts, not the loudest
The quality of the engine depends on the prompts that drive it. We did not guess. A cup system compared an initial field of eight prompts in head to head rounds. The same criteria applied in every match. Clarity and actionability. Prose that stands up to executive scrutiny. Traceability to project sources. Signs of self checking. When results were close, the tie moved to the prompt that produced denser evidence with fewer words. Decisions were recorded to leave an audit trail for later review. The method sounds like tournament language, yet its purpose is serious. It prevents fashionable prompts from slipping in, and it selects for durable performance.
The winning trio and why they work together
Three prompts remained. The first, The Human Method, reconstructs the leader’s implicit way of working. It asks how decisions are actually made under pressure, which trade offs recur, which questions signal quality, and which weak signals have proven useful before. It quotes and dates its sources, and it proposes gate questions to block weak reasoning before it spreads.
The second, Decode Your Method, acts as an excavator. It mines prior documents, calendars and notes to surface recurring tasks and patterns. It names them in operational language, links them to files, and proposes small routines that can be tested the same day. It is the most productive of the three because it converts past work into near term leverage.
The third, Analyze Me, builds a narrative profile that connects documented strengths to situations. It does not flatter. It anchors claims in artefacts, then sketches short learning paths with microsteps. The trio holds portfolio distance. One clarifies method, one exposes patterns, one writes a profile you can use in real conversations. Together they cover how, what and where to apply effort without stepping on one another.
Modern tooling that helps and leadership that does the work
Project memory keeps context across sessions and weeks, so the model does not start from zero each morning. Workspace and profile memory retain stable preferences. Deep research stitches files to questions and returns citations rather than vague references. These capabilities reduce friction. They do not remove the need for judgment. Framing the problem, naming the decision, and setting the guardrails remain squarely with the leader. The tools increase reach. The leader sets direction.
What changes when microactions compound
Within weeks the daily work would create a library of applied practice. Drafts, reviews, decisions and lessons learned are easy to find because they sit where they were made. The tutor’s questions sharpen the thinking muscles that matter most, such as counterexamples, second order effects, and the cost of delay. Coaching conversations would become faster and more concrete because they draw on what has been done, not on what might be done.
The preparation for day one in the new role is no longer a promise. It is already in motion.
Risk named early is risk reduced
Hallucinations still occur, though less often with stronger models. Bias remains a real concern. Privacy and data handling have to be explicit. The project addresses this by limiting sources to internal material during selection and by making data boundaries clear in every prompt. The goal is not to remove risk. It is to make it visible and managed in the same channel as the work.
Why experienced leaders have an edge with AI
The strongest returns show up where language and judgment are already refined. Rich vocabulary, fast sense making and a habit of asking for proof translate directly into better model behavior. You guide longer threads of thought, you read for substance before style, and you resist cheap certainty. The model then follows your line, produces cleaner text, and knows when to mark limits. That is not magic. It is leadership craft meeting a competent partner.
What to do next
Treat AI as an amplifier of your method and as a recorder of your proof. Place small actions inside your day, in one place, at one time. Choose prompts through a simple tournament rather than through taste. Keep the evidence near the words. Let microchanges build momentum you can point to. When the larger role begins, you will not be starting. You will be continuing.
