The truth about ‘winning with AI’

It’s that time of year again. Budgets, plans, a spreadsheet with a tab called 2027. And somewhere, a line that says AI.

Fair enough. But a lot of those lines get written in a slight panic. Someone on LinkedIn has an agent running their whole sales team, a competitor saves a day a week. Suddenly the pressure you feel to ‘win with AI’ leads to your ‘Ten AI Goals for 2027’ heading in 10 different directions.

I’d start somewhere quieter. Forget the noise. Look at your own organisation.

Three buckets

When we build an AI roadmap with a client, we sort every opportunity into one of three buckets. Educate: people learning to use AI in their own work. Automate: AI built into workflows that run across teams. Innovate: doing things you simply couldn’t do before.

Then you put your plans into those buckets and look at the shape. For one client, year one came out at 40% educate, 40% automate and 20% innovate. They were a bit surprised by how little innovation was in there. I wasn’t. A plan doesn’t have to do everything. It has to know what it’s doing.

Ask the right questions

For a while I framed all of this as ambition. How ambitious do you want to be?

Then it clicked. That’s the wrong question. You can be extremely ambitious inside one bucket. Getting a whole company properly confident with AI within a year is a serious goal. It just isn’t innovation.

So now I ask about focus. Where does your effort go this year? Where doesn’t it? Saying “not yet” to innovation is completely fine, as long as you know you’re saying it. That third bucket is where most of the value sits. If you leave it on the table, do it on purpose.

Start where you actually are

The right focus depends on your starting point, and this is where honesty helps. If most of your people haven’t gone further than asking a chatbot to tidy up an email, put your weight on ‘educate’. Automated workflows on top of shaky skills and messy data don’t hold. Once people use AI with confidence, the weight shifts to ‘automate’.

At one client we agreed that the goal for year one was simply to get everybody educated. Every initiative that year was tested against that one sentence. It made a lot of decisions much easier.

Measure each bucket on its own terms

‘Educate’ is mostly about adoption, with a bit of time saved. ‘Automate’ is about efficiency, and that’s easier to measure, because you’re looking at a workflow instead of a person. ‘Innovate’ is about impact. New value, new revenue, a different way of serving your customers.

Ask an ‘educate’ initiative for revenue impact and it will fail every time. Judge an ‘innovate’ initiative on adoption numbers and you’ll sell it short.

Before it goes in the plan

Four questions I’d put on the table:

• Where are we today, honestly?
• Which bucket gets most of our effort next year?
• What are we knowingly leaving for later?
• How do we measure each initiative, given its bucket?

‘Winning with AI’ means something different in every organisation. And most of them are no further along than yours.

Keep an eye on what’s happening out there, sure. Just don’t let the noise write your plan for you. That one’s still yours to make.