Why implementing AI in a company takes several months – and why that's a good thing
- Monika Kotus
- Mar 10
- 4 min read
When companies and organisations tell me they want to implement AI, the first thing that comes to mind is usually a training. A two-day intensive workshop, lots of material, lots of tools – and done. I understand the logic. A training is something concrete. It has a date, an agenda, an end.
The problem is that the results rarely stick after two days.
Over the past several months, I've been working with organisations in a different model – multi-month projects where we meet regularly, every week or every two weeks, for two to three hours. And I see a clear difference. Not just in how much people have learned – but in how they actually use these tools in their day-to-day work.
1. After two days of intensive training, people are overwhelmed
I don't mean this critically. It's just how the brain works.
When you show someone ChatGPT, Gemini, NotebookLM, Claude, Custom GPTs, Make and a handful of other tools within 16 hours – they walk out with a full head and are barely able to apply any of it. Not because they're not capable. Because that's simply not how learning works.
A model where we cover two or three tools per session – or work through one specific process together – gives much better results. Between sessions, there's time to test things. To break something. To come back with a question. And that question is the moment where something actually gets embedded.
2. The real questions come up only after some time has passed
One of the people I work with sent me a message a few weeks after our first session together: "Monika, I set up the assistant but it's not responding the way it should."
We sat down together and went through the instructions. The assistant was almost right – it was missing one piece of context and a few clarifications in the system prompt. Something I spot immediately after years of working with these tools. Something that's genuinely hard to catch when you're just starting out.
Those questions can't be asked during a workshop. Because at that point, you don't yet know you have them. They only appear when you sit down to work and something doesn't behave the way it was supposed to.
3. AI is only as good as the context you give it
There's another reason why several months of working together make a significant difference.
AI doesn't know who you are by default. It doesn't know your organisation, your way of writing, your processes, or your clients. That's why one of the first things we do in my projects is build context.
We gather documents – process descriptions, communication materials, FAQ, examples of good responses, and the way the organisation writes and speaks. We organise all of it in one place. This takes a few hours, sometimes a few sessions. But once it's done, AI genuinely adapts to the organisation. It doesn't generate responses that "sound fine". It generates responses that sound like it.
And then it starts working properly – faster proposal preparation, handling recurring questions, writing reports in the right style and tone. The model doesn't have to guess who you are, because it already knows.
This can't be built in two days during a workshop. Just gathering and organising those materials takes time. But once it's in place, every subsequent month is simply more productive than the last.
4. During implementation, there always needs to be a human checking the output
There's a lot of talk right now about automation and AI agents. And I understand the excitement. But I also see what happens when organisations move too quickly into full automation without a proper foundation.
In the projects I run, there's always a review loop – a person who looks at the AI's output and says: this works the way I want it to, or: something's off here, this needs adjusting. That's how we know where instructions need refining, where one tool should be swapped for another, what can be improved.
This isn't a weakness in the implementation. It's just the sensible approach – especially at the beginning, when the model is still learning the specifics of the organisation.
5. After a few months, it's not just the work that changes – it's the mindset
What I appreciate most about multi-month projects isn't the number of tools deployed. It's the fact that the organisation stops being afraid of this technology.
People simply start using AI in their work – not because they were told to, not because "you have to keep up with the times", but because they see it working and saving them time on specific tasks. That's the moment when implementation actually means something.
I work with organisations both in shorter projects and workshops – when the goal is a quick overview and first steps – and in multi-month projects where we go deeper. If you're wondering which approach fits your organisation, get in touch. We'll figure out together what makes sense for you.
