Softcat
Work out which AI ideas are worth funding
AI strategy finds where AI could solve a worthwhile customer or operational problem, then checks the value, feasibility, data, cost and risk. The result is a prioritised plan that says what to test first, what needs to be in place, and which ideas shouldn't go ahead.
When to call us
Too many ideas, no clear order
If AI ideas are piling up with no fair way to choose, a strategy decides what's worth funding first.
Everyone has ideas for AI, and nobody can say which are worth funding.
There are pilots in several teams, with no shared owner, architecture or order.
Competitors are adding AI features, and you need to know which customer problems are real.
You've been asked for an AI budget with no credible cost, benefit or point to stop.
AI projects are starting before anyone has looked at the data or the risk.
Why it matters
Most businesses haven't worked out where AI fits
Finding the right use cases is the hardest part, and few organisations have scaled AI beyond one or two teams.
-
71%
of UK businesses surveyed said they hadn't identified a use for AI, the most common thing stopping them adopting it.
Source: DSIT, AI Adoption Research, January 2026, opens external site
-
37%
of organisations surveyed worldwide use AI only at a surface level, with little or no change to how they work.
Source: Deloitte, State of AI in the Enterprise, January 2026, opens external site
-
22%
of organisations surveyed have successfully scaled AI across several business units.
Source: Gartner, AI survey, September 2026, opens external site
What we do
-
Collect the ideas
From leadership, product, operations and the people doing the work, plus the problems customers raise most.
-
Score them in the open
Value, feasibility, data readiness, integration effort, risk and running cost. You see the scoring, including for the ideas that don't make it.
-
Test the riskiest assumption
A quick technical spike or prototype on the top idea, so the plan rests on evidence rather than a demo.
-
Agree the order
A sequenced roadmap with owners, costs, what has to be in place first, and the governance each idea needs.
Worth knowing
A good AI strategy says no
Three things worth knowing before you start.
A strategy isn't a list of tools
Tools come last. The problem, the data, the process and the cost come first.
The biggest model isn't always the right one
A smaller model, a search over your own content, a few rules or ordinary software can be cheaper, faster and safer for the task.
Some ideas should stop
A good method parks ideas where the data, value or risk doesn't add up, and says why.
Self-check
Should this use AI?
Five questions to ask about any AI idea before you fund it.
- What problem does it solve, and how do you measure that problem today?
- Would a simpler fix, like better search, a form or a rule, solve most of it?
- Do you have the data it needs, and are you allowed to use it this way?
- What happens when it gets something wrong, and who notices?
- What will it cost to run each month once it's live, not just to build?
Questions
The things people usually ask
Plain answers about AI strategy and roadmapping, what's involved and what we need from you. Anything else, just ask.
Do we need clean data before we start?
No. We check whether each idea's data is good enough, which is a much smaller job than fixing all of it first.
Will you push us towards Claude?
We're an Anthropic partner and we build with Claude a lot, but we choose the model for the task. Sometimes the answer isn't a language model at all.
What's the difference between this and Uplift?
Uplift is a fixed-price version focused on your product. AI strategy work can cover the whole business, including operations.
