Four weeks into my new role at Collab365, I had to say something I thought decades of experience should have made unnecessary.
"I genuinely don't know how to do that."
I had been migrating posts from our old system to the new one. One wrong move could affect customer access or leave work half-finished. I could not see what was complete, what had failed or what needed checking.
I have spent much of my career in management accounting. I am used to controls, reconciliations and knowing where a number came from.
The new workflow gave me none of that confidence.
The useful skill was not pretending
My first instinct was to keep wrestling with it quietly.
That would have protected my pride and increased the risk to the business.
Instead, I asked Mark for a simple flag system:
- green for done;
- yellow for review;
- red for a problem.
Once the state became visible, the panic dropped. I could work through the migration without holding the entire process in my head.
Mark built the change. I did not suddenly become a software engineer. I used the experience I already had to explain what a controlled process needed.
AI skills are not one thing
The original version of this article used dramatic survey percentages to claim almost everyone had an AI skills gap. Those figures came from different samples and did not prove the personal lesson.
The UK government's AI Labour Market Survey does support a more careful point: employers report difficulty finding a mix of technical and non-technical AI capabilities.
That does not mean every professional has been reset to zero.
My accounting judgement did not disappear because the tool was unfamiliar. It helped me spot the missing control. Mark's technical ability did not tell him where I was getting stuck until I described the problem.
The new skill sat between us: exposing uncertainty early enough to fix the process.
What I do now
When a new AI-assisted workflow lands on my desk, I ask:
- What is the outcome?
- What can this change or damage?
- How do I know which state it is in?
- Where is the source or audit trail?
- Who checks an exception?
- How do we undo a mistake?
"I don't know" is not the end of that conversation. It is the point where a safe one can begin.
If you are trying to use Copilot without pretending every workflow is obvious, join Microsoft Copilot Adopters.