The build got faster. The problem did not get simpler
Generative AI has compressed the timeline for building data tools dramatically. Extraction pipelines, classification models, document processing workflows: projects that took the better part of a year using traditional machine learning techniques can now be delivered in weeks. That is a genuine and significant shift.
But the acceleration applies to a specific part of the process. It applies to the building. It does not apply to the thinking that comes before it: understanding what problem you are actually trying to solve, whether the problem is worth solving with technology at all, and what shape the solution needs to take to be useful in practice. That part is no faster, no cheaper, and no less important than it was before GenAI arrived.
Faster tools, same wrong answers
When the cost of building drops, the cost of building the wrong thing drops too. That sounds like a benefit, but it creates a real risk. Organisations can now move quickly from a loosely defined idea to a working prototype, and the speed of delivery can mask the fact that nobody stopped to ask whether the idea was right.
A well-built tool that solves the wrong problem is still a failed project. The difference is that you find out faster, and the temptation to keep iterating on something that was misconceived from the start is stronger when each iteration feels cheap. The discipline of framing the problem correctly before writing any code matters more now, not less.
The question behind the question
In practice, the stated requirement is rarely the real one. A client may ask for a search tool, but what they actually need is a way to understand whether their data is being captured properly in the first place. Another may ask for a pricing model, but the underlying issue is that the parameters driving price are not being surfaced early enough in their sales process.
Reaching the right problem requires experience, direct conversation, and a willingness to challenge the brief. It is not something that a model can generate. It comes from working closely with teams inside the business, understanding how they actually operate, and recognising the patterns that connect what different organisations in the same sector are experiencing.
Seeing patterns across an industry
One of the most valuable things an external partner can bring is perspective across a sector. When you work with multiple organisations in the same industry, you see the same problems surfacing independently. You can recognise when two separate clients are converging on the same question before either of them knows the other is asking it.
That kind of pattern recognition does not exist inside any single company. It requires breadth of engagement and enough depth in each relationship to understand what is really going on beneath the surface. It is also the kind of insight that no off-the-shelf AI product can offer, because it depends on context, relationships and accumulated judgement rather than processing power.
Technical risk is only half the picture
GenAI projects still carry real technical risks. Scalability, accuracy, reliability, security: these need proper engineering and they should not be underestimated. But sitting alongside the technical risks, and frequently given less attention, are the strategic risks. Building the wrong thing. Missing what your sector is quietly converging on. Solving a problem that did not need new technology in the first place.
The organisations that get the most from AI are not necessarily the ones with the most advanced technology. They are the ones that invest the time in understanding the problem before they start building, and that work with partners who bring enough sector context to challenge assumptions early.
Where to focus your GenAI efforts
Before commissioning a GenAI project, it is worth asking a few direct questions. Is the problem clearly defined, or are we building something because we can? Have we spoken to the people who will use this tool about what they actually need? And do we have a partner who understands our industry well enough to tell us when we are asking the wrong question?
Red Olive works with organisations across insurance, housing, financial services and other sectors to solve data and AI problems. The technology is part of what we do, but it is not the starting point. We start with the problem, and we bring enough cross-sector experience to help clients see what they cannot see from the inside.
