Cutting twenty candidate AI use cases down to the six worth building
Three digital product teams inside a global energy technology business had twenty proposed agentic AI applications, no shared way to compare them, and no budget to build all of them.
Each team assessed its own ideas on its own criteria — some on technical elegance, some on customer anecdote. Nothing was comparable, so prioritisation defaulted to whoever argued loudest, and the organisation could not tell an experiment from a product.
We scored every candidate on two independent axes at once: commercial evidence of demand, from customer interviews across five regions, and engineering feasibility, assessed with the teams who would build it. Anything strong on only one axis was explicitly parked rather than quietly funded.
- —Twenty-use-case inventory with standardised scoring model
- —Customer demand interviews across five regions
- —Feasibility assessment with each product engineering team
- —Revenue potential model per prioritised use case
- —Sequenced build roadmap with parked-item rationale
Six use cases carried both commercial evidence and a credible build path, with €3.4M of revenue potential attached per project. The scoring model became the standing mechanism the teams used for subsequent proposals.
Have a problem that looks like this one?
Start with a 45-minute discovery call. No pitch, no deck — we align on the problem and tell you honestly whether we are the right team for it.