The problem isn't your prompts
Going in-house with AI gets you speed. It rarely gets you quality. The gap is context, not prompting, and it's fixable.
Blog
Thoughts on brand, AI, and keeping creative humans in the loop.
A PDF can't tell you what's missing
Most brands think they're done because they have a guidelines document. For AI to work like a creative director, that document is maybe a third of the job.
On-brand isn't the same as good
Checking work against a brand's own rules misses two things: work that follows the rules and is still bad, and work that breaks them on purpose and is right to. Both need design knowledge.
Why brand context matters more than better prompts
Most teams assume their AI output problems are prompting problems. They're not. They're context problems — and fixing them means thinking differently about what your AI knows.
Your clients are going in-house. Good.
The shift to in-house AI looks like a threat to agencies. It isn't. The agencies that win are the ones that help their clients do it well.
Everyone has the same tools. So why does your work look like everyone else's?
When every team is using the same models, the tools stop being the differentiator. The only thing left that separates you is judgement.
Brand guidelines were written for the wrong reader
Guidelines were built to stop humans misusing a logo. Now a machine is reading them, and it needs almost the opposite of what they were designed to give.
AI that waits for a brief is wasting half its value
Most teams use AI as something you ask. The bigger prize is something that notices, suggests, and shapes a plan before you've asked anything at all.
The details a creative director notices, and a tool skips
Line height, headline capitalisation, whether a title ends in a full stop. The small things that separate work that's fine from work that's finished.