Before You Buy an AI Tool, Ask Three Questions

AI is extraordinarily capable, but it doesn't arrive in an organisation and create value by itself. It needs something to work with. It needs our data, documents, knowledge, systems and processes. It needs organisational context. And it needs us to be able to guide it.

1. What are we connecting it to?

First, we need to understand what we're connecting it to. Before AI can reason across our organisational knowledge, we need to understand what that knowledge actually is. Where does it live? Which systems contain it? What's sitting in documents and spreadsheets? What's structured, what's unstructured, and what still exists largely in people's heads?

We don't need everything to be perfect. But we do need to understand what we're giving AI to work with.

2. Are we automating the future - or the past?

We need to decide what we're asking it to do. Are we using AI to support better ways of working, or simply applying extraordinary new technology to processes designed around the limitations of the past? This is perhaps one of my favourite questions.

Many organisational processes were designed around the technology available five, ten, or twenty years ago - and those processes became "how we do things."

But shouldn't we now be designing our work based on the technology of the future, rather than automating the past?

3. Where are we directing it?

Once we understand what AI has to work with and have thought seriously about what we're asking it to do, there's one final question - we need to ask where we're directing it. What does meaningful value from AI actually look like for this organisation? Where do we actually want it to take us?

Answering these questions presents a very different perspective than being impressed by a list of features.

The tool isn't the first decision

AI needs us to have decided what we're connecting it to, what we're asking it to do, and where we're directing it.