Iterative AI Creative Partner
Brief, critique, and refine AI creative until the concept becomes usable
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 4
- Confidence
- 98%
Use an image-capable AI as a creative partner by giving it the same structured direction and revision cycle that a human designer would receive. Begin with the campaign proposition, audience, formats, and desired action. Generate a first round, inspect how the model interpreted every phrase, and identify specific failures such as illegible text, irrelevant imagery, or a misunderstood pun. Revise the brief by explaining ambiguous language, prescribing the scene, and supplying stylistic references. Continue producing and critiquing rounds until the underlying concept is strong enough for refinement or handoff. The mechanism works because each revision reduces interpretive freedom while preserving rapid experimentation, allowing marketers to explore more creative directions before committing production resources.
Origin
Extracted from Marketing Against The Grain through Kieran Flanagan's Zapier billboard experiment using ChatGPT, Bing, and image generation.
Core principles
- 01Treat AI like a creative collaborator, not a one-shot generator
- 02Specific briefs produce stronger creative
- 03Explain wordplay and ambiguous concepts explicitly
- 04Use successive revisions to correct interpretation and execution
How to run it
- 1
Frame the campaign
State the proposition, audience, desired response, and every asset format required. Include the exact copy that must appear.
Pro tip Ask for several placements, such as a billboard, bus, and digital ad, to expose weaknesses early.
Watch out A clever line without context may be interpreted literally.
- 2
Generate the first round
Let the AI translate the brief into initial concepts, then assess composition, imagery, copy, and brand fit separately.
Pro tip Treat the output as a concept board rather than finished production art.
Watch out Do not accept visual novelty as evidence that the communication works.
- 3
Diagnose interpretation errors
Identify exactly what the model misunderstood and explain the intended meaning. Clarify puns, product references, and the role of each visual element.
Pro tip Describe the scene as observable actions and objects.
Watch out Blaming the model without improving the brief produces repeated failures.
- 4
Constrain and iterate
Add relevant references, remove unnecessary elements, and request another revision. Repeat until the direction is usable or ready for a human designer.
Pro tip Change one major variable at a time so you can see what improved the result.
Watch out Image models may still render logos and text incorrectly.
In the wild
Kieran asked AI to turn “Zapier automates the sheet out of work” into billboard, bus, and digital concepts. Early versions confused Google Sheets with bedsheets and produced poor text. He then described sheets floating in the sky and a happy user pressing automate, producing a much more relevant concept in roughly half an hour.
→ The process moved a campaign direction from an abstract tagline to a near-usable visual concept in hours rather than weeks.
Common mistakes
Using a vague brief
The model fills missing context with literal or irrelevant interpretations, especially when the idea depends on wordplay.
Expecting a perfect first revision
AI creative requires briefing, review, and multiple rounds just as human creative work does.
Ignoring production limitations
Strong concepts can still contain malformed text, logos, or layout details that require downstream correction.
Is it for you?
Best for
Marketers who have a campaign concept but need rapid visual exploration across several formats.
Not ideal for
Teams expecting polished, brand-compliant production assets from a vague one-line prompt.
From the transcript
“So the more specific you can be, it's actually phenomenal.”
“AI is a good creative partner if you brief it, keep iterated. Just like you would with humans, by the way.”
“Which is going from weeks to hours.”
From the episode
6 AI Growth Hacks Top Marketers Don't Want You To Know (#167)