Generate-Then-Edit Creative Workflow
Pair AI’s first draft with intuitive human editing controls.
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 97%
Design creative AI as a two-part system rather than a generation demo. First, use AI to remove the blank page by pre-assembling a credible initial artifact. Second, give the human an intuitive editing environment where taste, corrections, brand requirements, and situational judgment can shape the result. The editing experience is not secondary: poor revision controls can erase the value of excellent generation and cause users to leave. Teams should test the entire path from initial request through final customization, reducing both prompting effort and editing friction. The result combines AI speed with the control users need to claim and trust the finished work.
Origin
The panel contrasted the strength of AI creation with weak editing experiences, while Grant Lee explained that Gamma had built editing before adding AI generation.
Core principles
- 01Creation and editing are separate product problems.
- 02AI should eliminate the blank page without eliminating human taste.
- 03A generated draft is valuable only when users can control it.
- 04Existing editing foundations can become an AI product advantage.
How to run it
- 1
Define the finished artifact
Identify what users must be able to change before they consider an output complete and usable.
Pro tip Study real revision histories.
Watch out A visually impressive draft may still be operationally unusable.
- 2
Generate the starting point
Use AI to assemble enough structure and content to eliminate blank-page effort.
Pro tip Optimize for editability as well as appearance.
Watch out Overproducing details can make correction harder.
- 3
Expose intuitive controls
Let users revise text, structure, media, style, and sequencing without rebuilding the artifact.
Pro tip Use familiar direct-manipulation patterns where possible.
Watch out Do not force every correction through another prompt.
- 4
Preserve human taste
Make human review and customization a normal part of the workflow rather than a failure state.
Pro tip Offer strong defaults without locking them.
Watch out Automation that suppresses judgment can produce generic or off-brand work.
- 5
Test the complete journey
Measure whether users can move from request to final deliverable, not merely whether generation succeeds.
Pro tip Include editing completion in activation metrics.
Watch out Generation benchmarks alone hide the main source of churn.
In the wild
Gamma first developed an easier editing environment for people without design backgrounds. It later added AI onboarding and generation so users could experience an immediate first draft while retaining the ability to edit the resulting deck.
→ The combined system lowered the initial learning curve without removing human control.
Common mistakes
Treating editing as an afterthought
Users abandon generated artifacts when routine corrections become harder than creating the work manually.
Making prompts the only control
Repeatedly describing small edits in natural language is often slower and less precise than direct manipulation.
Is it for you?
Best for
It is best for AI products that create decks, videos, images, documents, or other subjective artifacts.
Not ideal for
It is not ideal for fully automated outputs that require no subjective review or modification.
From the transcript
“I think one of the biggest problems historically for AI creative tools isn't actually the creation of the thing, it's the editing of the thing.”
“Now that the AI can pre-assemble and give you that first draft, you as the human can still go in and do all the editing…”
From the episode
The AI Workflow That Lets 50 People Do the Work of 500 ($2B Founder Reveals)