A large organisation's leadership team came to us with a familiar problem. They were hearing plenty about AI. What they did not have was an agreed view of where AI should start in their own business.
So we ran what we call an AI Futures Day. One day, on site, built so the people who actually run the business generated, sorted, scored and ranked the opportunities themselves. By the close they had a ranked top three, an owner and a date against each, and a working prototype of their first choice, built live in twenty minutes.
Any leadership team could run the same day.
The usual starting point
Most AI conversations start in the wrong place. They start with the tool, then go looking for a problem it can solve.
The other trap is the loudest idea in the room. A team under pressure to show progress can easily get pushed toward whatever the most senior person suggests.
The day was designed to deliver three things:
- A view of where AI could genuinely help, grounded in how work gets done today
- A way to judge each idea on value for money rather than enthusiasm
- A decision with owners attached, rather than another list
Four design rules
Write before you talk. Every contribution starts privately, on each person's own phone, before any discussion. This stops the most senior voice from anchoring everyone else.
People go first. No AI generated material enters the room until people have produced their own ideas. We opened with a simple prompt: describe a bad day at work, in specifics. Then came a silent inventory of pain points, typed on phones and shown on the screen one card at a time, word for word.
Show every score. When ideas are scored for value and effort, every person's score goes on the screen, never just the average. The conversation then goes to where the room disagrees most, because that is where the useful information sits.
Let structure carry trust. If the facilitator is new to the group, the structure does the work. Private scoring, anonymous reveals, and clear rules about what will and will not be discussed.
Phones instead of sticky notes
We gathered everything through simple online questionnaires that ran on any phone.
Before the day there were two sets. A named set went to the leaders in the room, tailored to what each person owns. A second, anonymous set was open to anyone in the wider workforce. It asked only which team the person worked in, so people could be candid.
On the day, one QR code put every phone into the session. As the slides moved on, each phone switched to the matching activity: adding pain points, building on someone else's idea, scoring, ranking and so on. Nobody scanned twice.
Every card, score and comment was captured as it happened, so the write up came straight from the room's own entries before anyone left.
How the day runs
- Set the ground rules. Declare up front which AI tools will be used and where data will and will not go. That answers the governance question before anyone has to ask it.
- Surface the work. People group their own pain point cards into themes. The facilitator keeps hands off, so the themes belong to the room.
- Generate candidates. Small groups turn the themes into candidate AI uses on a common template. Then a silent gallery walk lets everyone add notes to every idea.
- Score and rank. Everyone scores value and effort privately, the group talks about the biggest disagreements, then everyone scores again. The leaders' scores sit beside the wider workforce's answers, and the list is forced into a top three with no ties allowed.
- Build and commit. Build a rough working prototype of the number one choice with a twenty minute hard stop. Run a short premortem on the top three, imagining each has failed and asking why. The most senior leader then names an owner and a date against each.
What comes out
- A decision. A ranked top three, each with a named owner and a first date.
- Something real. A prototype the team can see, question and test, which does more than any slide deck to make AI feel practical.
- A ranking that holds up. Because every score was visible and the disagreements were aired, the priorities can stand up to scrutiny from whoever funds them.
- A voice for everyone else. The anonymous questionnaire brings in the people who do the work but were not in the room.
What we learnt
- The strongest ideas came from the pain point cards, never from a list of AI capabilities.
- Declaring the tools and the data boundaries first removed the biggest source of hesitation.
- Disagreement is useful information. Discussing only the widest spreads gave a better ranking in less time than open debate.
You do not need custom software to try this. A free form tool for private input, a whiteboard for themes and a spreadsheet for the scores will get you most of the way.
Try this. Ask each person on your leadership team to privately write down the three most tedious tasks in their week, before you discuss AI at all. Compare the lists and you have the start of your shortlist.