How I found my 5 highest-leverage AI use cases in 90 minutes (and what surprised me)
This is your fastest and simplest way to start with AI
I am a Global Product Manager. I manage a half-billion dollar crop protection portfolio across Europe, Latin America and Asia. I make long-cycle strategic decisions (country selection, active ingredient positioning, COGS structure, regulatory strategy) with 6-month approval timelines and feedback loops that stretch years.
I also built an AI use-case discovery method for commercial leaders.
Last week I ran it on myself.
Why I built the method
Most AI adoption frameworks fail commercial leaders for the same reason: they start with the technology and work backward to the use case.
“Here are 50 things AI can do. Which ones apply to you?”
That is the wrong sequence. It produces wish lists. Leaders end up chasing what looks most impressive in a meeting, and nothing ships.
The method I built starts from the opposite direction. It starts from your operating reality: your decisions, your pains, your data. It surfaces the use cases already embedded in your daily work. The ones you stopped seeing because you adapted to them.
The output is a ranked shortlist of 5, scored by leverage, tagged with AI shape and effort, and anchored to one concrete first move.
The five phases
Phase 1: Context capture (10 minutes)
Before anything else, I capture five things: role and what you actually own day-to-day (not the job description), P&L scope, sector and sub-sector, the 3-5 decisions made most often and their frequency, and the 3 biggest pains in your operating rhythm.
This is not small talk. It is the foundation for everything that follows. Every lens in Phase 2 is primed with examples from your specific sector. Generic examples are useless. A crop protection example lands differently than a consumer goods one.
When I ran this on myself, I described my three biggest pains as: scattered and unreliable data across too many disconnected sources; constant repurposing of the same strategic content for different stakeholders; and energy-intensive pre-alignment across functions before any decision can move.
I had said these things before. In meetings, in feedback sessions, in frustration. What I had never done was treat them as inputs to a structured discovery process.
Phase 2: The Six Lenses (25 minutes)
This is the heart of the method. Six lenses, two passes.
The first pass covers operational drain:
Repetition: what you do over and over
Time: what eats your calendar
Error: where manual mistakes live
Friction: where you think “this should be easier”
The second pass covers strategic blockers:
Aspiration: what you wish you knew but don’t have time to find out
Documentation: what someone explains to every new joiner because it was never written down
The lenses are not questions. They are priming sequences. For each one, I first give 2-3 concrete examples from the leader’s sector. Then I ask. Then I probe: surface answers are not accepted.
When I asked myself the friction lens question, I described our internal knowledge fragmentation: dozens of disconnected SharePoints, PDF reports, PowerPoints with no owners, institutional knowledge that disappears when someone goes on leave or leaves the company. You always need a human to explain any document you receive.
Then I said something I hadn’t planned: “Maybe we should have an automated sequential workflow. I initiate, it goes to my partner, he confirms, the AI analyzes the input, then it passes to the next person.”
I had designed a use case in real time. That is what the lenses do: they surface thinking you already have but haven’t made explicit.
Phase 3: The Decision Map (20 minutes)
Every candidate from Phase 2 gets scored on three dimensions:
Value (1-5): If this improved by 80%, what would materially change?
Frequency (1-5): How often does it occur? Daily=5, Weekly=4, Monthly=3, Quarterly=2, Annually=1
Feasibility (1-5): How clean and available is the data you’d need?
Leverage Score = Value × Frequency × Feasibility. Maximum 125.
Before scoring, I ask one question: are you optimizing for Business impact, Personal value, or Both? This is not a rhetorical question. A senior leader who cannot move the corporate P&L but can reclaim a week of her life every month should score for Personal honestly. Both are legitimate.
The scoring is where the method earns its value. In the math, yes, but more in the honest calibration it forces.
The most common mismatch: leaders score Value 5 and Feasibility 1 on the same use case. The aspiration is real but the data isn’t there. The formula makes that visible. A use case with Value 5, Frequency 3, Feasibility 1 scores 15. A use case with Value 3, Frequency 3, Feasibility 5 scores 45. The “less exciting” one wins.
That is exactly what happened to me.
Phase 4: AI Shape Tagging (15 minutes)
The top 10 candidates get tagged with an AI shape:
Assistance: AI drafts, human reviews and decides. Typical build: days.
Automation: AI runs end-to-end with safeguards. Typical build: weeks.
Analysis: AI surfaces patterns from data. Effort depends on data quality.
Generation: AI creates net-new output. Effort depends on quality-control needs.
The shape tag matters because it sets realistic expectations before anyone talks to IT or a vendor. “We want AI to help with market sizing” sounds concrete until you realize the data lives in three countries’ Excel files with no common structure. That is a months-long Analysis build, not a days-long Assistance task.
Phase 5: Shortlist and First Move (15 minutes)
The final output: ranked top 5, each with shape, effort, and one recommended first move.
The first-move recommendation is what separates this from a workshop output. The scoring tells you which one to build this week, and why. One of five strategic reasons: lowest implementation friction, highest visible win, unlocks the others, easiest to demonstrate ROI, or returns the most personal time.
What surprised me
My top 5 use cases where I can start now:
Async decision flow
Regulatory data queries
Monthly meeting input prep
Country/region request handling
Onboarding and ways-of-working
The recommended first move: regulatory data queries., as it is very fast and easy to automate. I receive 5-8 identical requests per month from the regulatory team, each taking 30 min of my time. So i found a way to free 4 hours in my diary!
I had lived with it so long it had become invisible. I stopped counting it as a problem. I just did it.
The method made it visible again and it sends a clear message. I had been undervaluing my own time on tasks that required zero judgment. Such tasks feel “small” and I just did it. The regulatory data queries felt administrative and boring.
Boring and automatable is exactly where you start with AI automation.
Who this is for
I built this method for commercial leaders managing complex portfolios: product managers, regional directors, heads of category, general managers. People who make strategic decisions but are also buried in operational grind. People who know AI should be part of their work but don’t know where to cut in.
The method takes 90 minutes. It requires no technical knowledge. It requires honesty: about where your data actually lives, about what is genuinely high-value versus what just feels important, about the difference between a political blocker and an information gap.
If you want to run it yourself, subscribe below. The skill file and activation guide arrive in your first email. Free for the first group testing it.
The discovery is the first step. The next is pressure-testing the chosen use case against data, governance, and team readiness.
But you have to know what you’re pressure-testing first.
Great insights! Thank you
love it!!!