how do we get started with ai?

ai tools can be bought on subscription today and be up and running before lunch. it is not the technology that stops you. it is everything that should have been in place before you even started talking about ai.

it is not the technology that fails

most ai projects do not die because the model is wrong or the tool is poor. they die in three other places. data that is scattered and inconsistent. processes that are not ready to be automated. and an organisation that has not decided what ai does to the people who work in it.

ai does not build anything new from scratch. it amplifies what is already there. or what is missing. that is why an ai project quickly becomes an x-ray of the company's real maturity.

the four questions we always ask first

1. data. where does your customer data live? how clean is it? and can it be accessed in a structured way by the systems that have to activate it? at a club, customer data typically lives in four places at once, in the ticketing system, the membership database, the webshop and the email tool, without any of them knowing it is the same person.
2. processes. which workflows do you actually want to automate? and what happens in the organisation when a human no longer has to carry them out?
3. people. what happens to the specialist who can see their job changing, and how do you help the colleague who can't find their own way into ai?
4. ownership. who owns the ai initiative commercially, who owns it technically, and who has to be able to say no to a project that does not create value?

when the four questions have clear answers, ai suddenly becomes the simplest part of the project. when they do not, ai becomes yet another system that costs time and money without moving the bottom line.

where the hours typically go in operations, and what needs to be in place before automation makes sense, is covered on the technology and digitalisation page.

the conversation nobody has

most of what is written about ai is about technology and use cases. that is the easy part. the hard part happens in the corridor, by the coffee machine and at the kitchen table in the evening.

the specialist who has spent 15 years getting good at a discipline can suddenly see an algorithm doing 60 % of the work. what should her role be in two years?

the 60-year-old colleague who has not played with chatgpt in his spare time quickly falls behind. how do you bring him along without him losing his footing and his confidence?

and then there is the junior who has just joined. most of the work she was supposed to learn her trade on is now done in 20 seconds. how do you make sure she still builds the judgement that makes her a good adviser in ten years?

these are not hr questions. they are strategic questions. because if you have no position on them, you lose employees before you get any benefit from ai. and you risk waking up in five years with an ai stack that is state of the art and an organisation that can no longer think for itself.