Most AI strategies follow a familiar shape. A handful of leadership interviews, a workshop, a long list of use cases, a slide deck. Then a few pilots that never quite connect to how the business runs.
The problem is not ambition or budget. It is information. The people who design the strategy are rarely the people who do the work, and the work is where AI either helps or does not.
The information problem
Ask a senior leader where time is lost and you will get a sensible answer: reporting, approvals, customer queries. Ask the person who runs the monthly report and you will hear something more specific. The data comes from three systems, one of them exports only to PDF, the numbers are re-keyed into a spreadsheet, and the whole thing waits two days for a sign-off.
That second answer is the one an engineer can build against. It names the tools, the handover, the manual step and the delay. It is also the answer that never reaches a strategy workshop.
What the research says about task selection
Controlled studies show AI can make knowledge work dramatically faster. Support agents resolved 34% more issues per hour when they were newer to the job (NBER, 2023). Professionals spent 40% less time on writing tasks (MIT News, 2023).
The same research carries a warning. In the Harvard Business School and BCG study, consultants using AI were faster and better on tasks inside AI's capability, and worse on a task outside it (Harvard Digital Data Design Institute). The researchers called this a "jagged frontier".
Put simply, the value of AI depends on choosing the right tasks. You cannot choose well without knowing what the tasks are.
Why interview everyone
Traditionally, talking to every employee was impossible. A consulting team might manage fifty interviews in a month. Organisations with hundreds or thousands of people had to rely on a sample, usually senior.
AI agents change the arithmetic. At Thirty Days, every employee has a private 30-minute conversation with an AI interviewer, in Arabic or English, at a time that suits them. Hundreds of interviews run in parallel, so discovery takes days rather than months.
Three things follow from that.
- The map is complete. Every department, including the ones that never make it into strategy meetings, describes its own workflows, tools and handovers.
- People are candid. Responses are anonymised in everything we deliver, and people are often more open with an AI than with an outside consultant.
- The evidence is quantified. When forty people independently mention the same approval delay, it stops being an anecdote and becomes a ranked priority.
From interviews to a ranked roadmap
The interviews produce thousands of signals: tools used, manual steps, waiting times, workarounds, ideas. We turn them into a map of how work flows between people and systems, and rank the processes where AI will return the most time and money.
That ranking drives everything that follows. Our forward-deployed engineers build against the top items, working on site with the people who described them. Our training programmes are written from the same interviews, so each team learns AI for the work it described.
What changes for leadership
Leaders still set direction. What changes is the quality of the evidence behind their decisions. Instead of a long list of plausible use cases, they see where hours are actually lost, how much it costs and which fixes are realistic in thirty days.
That is also how results get proven later. A process that was measured before AI can be measured after it.
If your organisation has an AI strategy that has not yet reached daily work, see how the thirty-day method works or talk to us.
