Align · Second stage of your AI journey
Use case prioritisation
Most organisations do not have an AI idea problem, they have an AI comparison problem. Somebody in HR wants a document summariser, somebody in finance wants forecasting, and there is no honest way to put those two things side by side. This engagement gives you that comparison in weeks rather than quarters.
A wish list is not a roadmap
Once people inside your organisation start paying attention to AI, ideas arrive from everywhere. That is a good sign, and it is exactly what the Bootstrap stage is designed to provoke. The difficulty comes next, when you have forty suggestions written as one-liners in a spreadsheet, all competing for a budget that will cover three of them.
What usually goes wrong at this point is predictable, and it is rarely about ambition:
- Ideas arrive too thin to judge. "Use AI for onboarding" could be a fortnight of work or a year of development, and nothing in that sentence tells you which is it.
- Value is scored by the person who proposed it. Everyone believes in their own idea, so everything comes back rated high, and the ranking collapses into a popularity contest.
- Feasibility is guessed at by people who have not built this before. Genuinely hard problems get waved through because they demo well, while something quite achievable gets dismissed as science fiction.
- Regulation turns up late. A use case sails to the top of the list, and then somebody notices it makes decisions about people, which puts it squarely in the high-risk provisions of the EU AI Act with obligations nobody costed.
- Nothing has a name against it. The roadmap is agreed and six weeks later it turns out no individual was ever accountable for any line of it.
The research on this is not flattering. According to MIT NANDA’s 2025 study, around 95% of enterprise AI pilots deliver no measurable impact on the P&L, and S&P Global found that 42% of companies abandoned most of their AI initiatives during the same year. Very little of that waste comes from choosing a hard problem. Most of it comes from choosing without a method.
Source: MIT NANDA, The GenAI Divide: State of AI in Business (2025); S&P Global (2025)
Fixed scope, fixed timeframe, with a decision at the end
Use case prioritisation is one of our Align engagements, which means it is (mostly) pre-defined rather than fully bespoke. You know before you start what happens, who needs to be involved, and what lands on the table at the end. That makes it straightforward to price, straightforward to approve, and hard to let drift.
It runs over three to six weeks, depending on how many functions you want to involve, and it has three movements: we collect the ideas properly, we assess them against metrics that can actually be verified and quanitified, and then each idea owner pitches the use cases to a roundtable of stakeholders, with us as moderators.
That last part matters more than it sounds. We could score your backlog in a back office and email you a ranked list, and it would be a perfectly defensible list that nobody in your organisation felt any ownership of. Instead, teams pitch their own use cases and defend them. By the time you decide what gets built, the people who will own those products are already in the room and already invested. You are not choosing work and then hunting for someone to do it.
Three phases over three to six weeks
Shaping (before the work starts)
A short workshop to agree who takes part, which parts of the business are in scope, and how the scoring dimensions should be weighted for you. A bank weighs regulatory exposure differently from a manufacturer, and the model should reflect that before anyone uses it.
Phase one: Collect
We gather candidate use cases through a structured intake that asks for the few things that make an idea assessable: who has the problem, what happens today, what "better" would look like, and where the information lives. We run working sessions with each contributing function rather than sending a form into the void, because the good ideas usually surface in conversation and the person with the best one rarely fills in surveys.
We also record what is already running, including the tools people adopted without asking. That is not an audit and nobody gets into trouble; it is simply useful to know where your organisation has already started.
Phase two: Assess
Every collected use case is then assessed on two sides.
On the value side we work with the proposing team to express the benefit as something measurable, whether that is hours returned, cost avoided, revenue influenced or risk reduced, together with how many people or transactions it touches, and how well it fits the direction the business is already heading.
On the feasibility side we look at:
- Data readiness. Does the information exist, can it be reached, and is it in a state anyone would trust?
- Pattern maturity. Is this a well-understood pattern we could build with confidence, or is it still a research problem?
- Integration surface. How many of your systems does it have to touch, and do those systems have a simple way in?
- Regulatory exposure. Where does it sit under the EU AI Act, and what else applies given your sector, whether that is GDPR, DORA or something specific to your regulator?
- Ownership. Is there a named person whose working life measurably improves if this succeeds? Use cases without one tend not to survive long.
We do the technical and regulatory assessment ourselves, and we do it as people who build these systems rather than people who read about them. A feasibility score is only worth something if the person giving it would be willing to deliver the product.
Phase three: Decide
Teams pitch their use cases to a room that includes your sponsor and us. Scores are applied live and visibly, disagreements happen in the open, and the weighting is adjusted if the room can justify it. You leave that session with a ranking your organisation has consented to.
We then turn the ranking into a sequenced roadmap and walk your sponsor through it.
Questions we get asked
How long does it take?
Three to six weeks, on average, from the first collection session to the pitches, plus a short shaping call beforehand. The range depends mostly on how many functions and teams you want represented.
Do you need access to our data?
No. We need conversations with the people who know where your data lives and what condition it is in, which is a much smaller ask and a much shorter approval path. These conversations will happen as use cases are discovered.
Can it run remotely?
Yes. The pitch session is the one part that genuinely works better with everyone in the same room, so we suggest doing that one in person where it is practical.
What if our ideas turn out to be bad?
Then you have learned that in weeks rather than after a year of building, which is a good outcome and a cheap one. In practice it is rarer than people fear; what usually happens is that the ideas are sound but the sequencing was wrong. At a very minimum, you'll get hints of where automation (not AI) can improve your team's processes.
Do we have to build with you afterwards?
No, and the deliverables are deliberately written so that you do not have to.
We already tried prioritising internally and it stalled. Why would this be different?
Usually because internal exercises struggle with two things: an independent feasibility view, and someone with the standing to make the trade-offs visible in a room where people disagree. Those are the two jobs we are there to do, and it's a much easier task for us to do this without the organisational politics and complex agendas that internal teams have to manage.
Ready to work on your AI use cases?
If you have a backlog of AI ideas and no comfortable way to rank them, that is a conversation worth having. Fill in the details and we'll send you the full description of this engagement, including what deliverables you should expect from us. If that aligns with your goals, then we can schedule a first call to work out the details.
