Forever Foundry Book a scoping call

Guide: Discovery

How to find AI use cases in your business: start with your people

Published by Scott Seaborn, Founder

The best AI opportunities are rarely on a vendor’s slide. They are in the everyday friction your people already know about.

Most lists of AI use cases start with the technology: what a model can do, what a vendor is selling, what a competitor has announced. The result is a list of things AI could do, not a list of problems worth solving.

The better starting point is the people who do the work. They already know where time is lost, where errors creep in and where they rely on workarounds. The task is to ask them properly and turn what they say into evidence.

Why top-down use-case lists disappoint

  • They reflect who was in the room. A workshop with senior managers captures their view of the business, not the friction three levels down.
  • They start with a solution. “Use AI for customer service” describes a technology, not a problem with a cost attached.
  • They miss the gaps between teams. Much of the waste sits in handoffs that no single department owns.
  • They lack a baseline. Without knowing what a problem costs today, nobody can show later that AI improved it.

Where the real opportunities sit

When you listen across an organisation, the same kinds of friction tend to appear.

  • Repeated manual work: re-keying data, copying between systems, building the same report every week.
  • Workarounds: spreadsheets and side processes that exist because the official system does not fit.
  • Waiting: work that stops until someone approves, checks or answers a question.
  • Rework: errors found late, orders corrected, documents requested twice.
  • Knowledge bottlenecks: work that queues for the one person who knows the answer.
  • Handoffs between teams: information that is lost, delayed or reinterpreted as it moves.

Some of these are good candidates for AI. Others are better solved by fixing a process, cleaning up data or connecting two systems. Finding the problem first lets you choose the right fix.

Questions to ask your people

Keep the questions about the work, not about AI. People describe problems more accurately than they guess at solutions.

  • Which tasks take more time than they should, and how often do you do them?
  • Where do you wait for information or a decision from someone else?
  • What do you work around because the system does not do what you need?
  • Where do mistakes usually get caught, and how late?
  • What do you know that colleagues regularly have to ask you for?
  • If you could change one thing about how this work gets done, what would it be?

Ask people to score how significant each issue is as well as describe it. Scores let you compare across teams; the explanations tell you what sits behind them.

Who to ask

Ask across functions, roles, sites and levels. The view from the shop floor, the office and the boardroom will differ, and the differences are informative. Track who has taken part, so you can see where coverage is thin rather than assume one team speaks for everyone.

Turn answers into use cases

Raw responses are not yet use cases. Group related issues into focus points, each described as a business problem with the evidence attached: who raised it, how often it occurs and what it costs today.

Then, for each focus point, ask what kind of change would solve it. AI may be the answer. So might process redesign, better data, an integration or training. Test whether the problem is real and significant before you design the solution.

Then prioritise

You will usually find more opportunities than you can fund. The next step is to rank them by business value and delivery complexity, with the people who know the work doing the scoring. Our guide to the value vs complexity matrix explains how to do that well.

A short checklist

  • Start with problems, not technologies.
  • Ask across functions, roles, sites and levels.
  • Ask about the work, and collect scores as well as explanations.
  • Track participation and flag gaps.
  • Group responses into evidence-backed focus points.
  • Choose the type of fix before choosing the technology.

AI Compass™ is how we do this: it gathers insight through AI-powered surveys, branded for your organisation and sent by you, and brings scores and explanations together into evidence-backed focus points. Your people then rank them through Slid3rs™, and leadership agrees what happens first in FlowRoom™.

Common questions

How do I identify AI use cases in my business?

Start with the problems your people experience, not with what AI can do. Ask staff across functions, roles and sites where time is lost, where they wait, what they work around and where errors occur, then group their answers into evidence-backed focus points before choosing any technology.

Who should be involved in finding AI opportunities?

The people closest to the work, across every function, role, site and level involved. Senior leaders set the decision, but the detail of where friction sits usually comes from the people doing the work.

Is AI always the right answer to the problems you find?

No. Some problems are better solved by redesigning a process, improving data, connecting systems or training. Identify the problem and its cost first, then choose the type of fix.

Related reading

Next

Find the friction worth solving.

Explore AI Compass™