Why AI projects get stuck in the pilot, and how to move on
The demo impressed the management team. Then nothing more happened. It is common for AI projects to stop after the pilot, and the reason is rarely the technology. Here are the most common causes and what you can do about them.
You started with the technology instead of the problem
Many pilots start with the question "what can we do with AI?". That makes fun demos but rarely anything that gets used. Start with a flow that takes a lot of time today instead, like sorting incoming cases, putting together quotes or looking up information across systems. Count how many hours it takes each week. Then you know what a solution is worth, and management has something to decide on.
The pilot was built on test data
A solution that works on ten hand picked examples can break on real data. Build early against the systems and data the solution will use in production, with the permissions that apply there. Then you see the problems while they are small.
The people who will use it weren't involved
When a solution is built far from the people who do the work, it rarely becomes part of their day. Let some of them test it every week and say what is missing. They are the ones who know where the exceptions are.
Nobody owns the solution after the pilot
A pilot often has a champion but no owner. Decide before you start who is responsible for the solution in production, who fixes it when it gets things wrong and how you measure that it saves time.
Risk and regulation came up too late
Data about customers and employees means GDPR applies, and the EU AI Act sets requirements for some kinds of solutions. If these questions come up only after the pilot, they can stop everything. Put them in the plan, so the solution is built to meet them.
How to take a solution all the way
- Map. Talk to management and the people who do the work, and go through systems and data.
- Prioritise. Rank the flows by how much time and money they can save, with estimated cost and risks. Let management decide what to build first.
- Build one flow at a time. Build against real data and improve the solution until it is used every day.
- Hand over. Teach the team to use, change and extend it, so the knowledge stays with you.
How we help
We put together an AI strategy management can decide on, then build the solutions with an AI consultant on site with you, from 1,200 SEK/h. You can also stop after the plan and build yourselves. Read more about AI strategy and AI consultants.