The company wants to "do something with AI"
The pressure has reached leadership, somebody said "we have to do something with AI" in a meeting, and now everybody is waiting for a plan. Before "which tool do we choose", the first good question is "what kind of intervention do we actually need": training, consulting or facilitation. They are not synonyms, and the wrong choice costs months.
We already have a simple grid for this, which we have used for years for any topic (not only AI): if people lack knowledge, you choose training; if the knowledge exists but the process that gathers it into a shared result is missing, you choose facilitation; if you need someone else's answer, with responsibility taken on it, you choose consulting. On AI, the grid looks like this:
| If… | Choose | Because |
|---|---|---|
| your people don't know how an LLM works or what an AI agent can do | training | the gap is in technical knowledge, and someone from outside has it |
| your people have licences and demos, but never reach a shared way of working | facilitation | the technical competence exists somewhere in the organisation; the process that aligns it is missing |
| you need a technical architecture or implementation, with responsibility taken on it | consulting / implementation | you are buying a system, not a conversation |
Most organisations misdiagnose their problem and buy the first column when they need the second.
"We bought licences. 10% of people use them."
The classic symptom of a subject treated as training when it is really a facilitation problem. The company has solved the knowledge part (people know, in theory, what an AI assistant is), but nobody has redrawn the workflow in which the tool would actually make sense. Without a personal reason to use it and without a process that integrates it, the licence remains a monthly cost nobody opens.
"Every department experiments on its own."
Shadow AI: people already use AI tools, unaligned and unsafely, because nobody has given them a shared direction. Leadership's reflex is to forbid. It doesn't work: the ban moves experimentation into an even less visible zone. What works is alignment: clear limits about what is allowed, what isn't, and who decides, followed by a process in which teams can compare and correct each other's ways of working. The right name for that is organisational alignment facilitation, closer to process than to training or consulting.
Where facilitation comes in, concretely
We are not an implementation firm: we don't write code and we don't configure agents. We complement the internal or external technical team: it delivers the system, we deliver the part that 70% of AI projects die from (aligning the leaders, redesigning workflows, real adoption in teams, direct work with the fear of change) which makes people smile in training and bypass the system afterwards. If you recognise any of the symptoms above, we have detailed what our AI adoption programme looks like, structured on exactly the three types of blockage: direction, competence, fear.
The complete comparison, with all the options (training, facilitation, consulting, team coaching), is in the dedicated article.
Not sure which column you fall into?
A free 30-minute diagnosis tells both of us whether the blockage is one of knowledge, process or something else.
Let's talk about AI adoption