Related question
Why did our AI results get worse after we standardized on a new tool?
Short answer
Because the old tool was compensating for weak prompting. When you use a system long enough it learns your patterns and quietly fills gaps in vague requests. A new tool does not know you yet, so it exposes prompting skill that was never actually there.
There are two different people inside this complaint and they need opposite responses. One built real skill and customized their old tool deliberately, and they need migration support rather than training. The other never learned to prompt at all and did not know it, because the old tool was covering for them. This answer is about the second one, and they are the larger group by a wide margin.
Here is how that trap works.
When you use an AI tool long enough, it picks up your patterns and compensates for incomplete prompts based on your history. The results feel good because the tool is doing work you do not realize it is doing for you.
Then your company standardizes. Everyone moves to a new platform. Suddenly your results feel flat, even broken. The new tool does not know you, and the conclusion is swift: the old tool was better.
That is rarely true. What the new tool has actually done is expose something the old one was hiding, which is your prompting skill.
The pattern plays out predictably. One person has used the same tool for two years and gets excellent output, and nobody knows exactly why. A colleague switches to the approved platform and gets mediocre results. Frustration sets in, the team concludes the standardized tool is inferior, and resistance spreads fast.
The real problem was never the tool. It was that nobody on the team was ever taught to prompt.