Observation
Short prompts, vague inputs, no background... and then disappointment with the output. The model isn't really the issue.
From the Field
I saw this with a simple task, drafting a policy update. The initial prompt was very basic, and the output came back generic and not particularly useful. When the same task was retried with more context, including the audience, the organisation's tone, and specific policy nuances, the quality improved significantly. Same tool, same task. Completely different result.
Why This Matters
AI fills in the gaps when you don't give it enough, and that's where accuracy starts to drop. The more relevant context you provide, the more usable the output becomes.
Field Takeaway
If the output isn't right, it's usually a context problem, not an AI problem.
Subscriptions open · New editions as published
THE MONTHLY FIELD NOTE
No noise. Just the useful bits.