Problem
Growth came in 24% under quarter goals while Adobe was losing share in a crowded GenAI field. Inside enterprise accounts, AI credits were sitting unused and renewal risk was rising.
GenAI Creative Production
Over 9–12 months, I led the design of Adobe's enterprise-scale GenAI creative production tools within Adobe Firefly. External: -24% growth on quarterly goals with Adobe losing market share. Internal: +60% of enterprise clients were not planning to renew due to unused AI credits. I identified the opportunity and designed a suite of tools enabling teams to produce assets at scale with brand consistency.
Enterprise Product Design, GenAI
An enterprise Firefly suite (bulk console, infinite canvas, workflow builder) that turns unused AI credits into shipped, on-brand assets.
- Role
- Product Design Lead
- Team
- AI researcher, solution architect, lead engineer, PM, UX researcher
- Timeframe
- 2024 to 2025, roughly 9 to 12 months
- What I owned
- Suite IA, bulk console flows, design-system patterns, vision work, and the MAX launch narrative.
Adobe Firefly Enterprise
Helping teams create at scale and on brand with GenAI.
How do you turn a powerful model into work that ships, on brand and at campaign scale, without a team rebuilding everything by hand?
The stakes
Growth pressure outside. Renewal risk from unused credits inside.
Solution
Create an enterprise Firefly suite for bulk generation, canvas work, and workflow automation, turning unused AI credits into shipped, on-brand campaign assets.
The squad
Six people, one of each discipline.
Discovery
We started with the people paying the bills.
With product and research partners, I interviewed our top ten enterprise clients and beta testers on one question: why a range this broad still went underused. Surveys sized what we heard.
Field research
Then we sat with the creative teams doing the work.
The accounts told us what to fix; their teams told us why.
- Coach
- Production lived outside the tool. Teams generated in Firefly, then rebuilt by hand in Photoshop and spreadsheets to hit campaign volume.
- Brand was a manual gate. Every asset got eyeballed against guidelines, so staying on brand never scaled.
- The work was collaborative, the tools were single player. Reviews, approvals, and handoffs happened in email and Slack, never on the canvas.
How we worked
Four moves, over nine to twelve months.
The gap
No surface for production at campaign scale.
Teams could generate one asset at a time, but had nowhere to produce a campaign: no bulk actions, no collaboration, no brand guardrails in one flow.
Experiment 01
Bulk actions, one job at a time.
The first bet ran single actions in bulk: remove background, resize, and recolor, across thousands of images at once.
Objective. Stand up a surface to test the GenAI APIs at scale, with real quality control.
Bulk console
The bulk console, from entry to job log.
Ten screens, one flow: configure the actions, upload thousands of files, preview on a sample, run the batch, track every job.
01. The way into the suite
Where teams enter creative production.
Results
3× adoption, 50% time saved.
Note
- 3× adoption. Enterprise clients validated the need, and usage grew across teams.
- 50% time saved. Versus the old Photoshop plus spreadsheet routine.
- A trade-off we named out loud. Stakeholders wanted a quick single-action release; users wanted more control. We shipped shallow to make GA, then shipped the depth next.
System
New patterns the whole suite could share.
Shared widgets for presets, custom sizes, background upload, color, and aspect ratios, reused across the console and the canvas.
Influence
Aligning leadership on the GenAI bet.
- Made the case to leadership for the GenAI investment, anchored on renewal risk and the adoption gap.
- Ran the workshops that turned the canvas and workflow-builder vision into a plan teams agreed to build.
Vision
Experiment 02: the suite, made collaborative.
Board
A board built for review.
Pinned comments, reviewer annotations, and brand-violation markers, right on the asset.
Quality
Brand guardrails, inline.
Brand checks run as you review, so off-brand work gets caught before approval.
Components
Dynamic, brand-aware components.
One component spins out its own variants, each shaped by brand rules and a prompt chain.
Prompts
Chaining prompts into a pipeline.
One pass: generate → mask → recolor → brand-check, end to end.
Workflow
A workflow builder on the canvas.
Drag in steps to apply brand, spin variations, or send for review, and the workflow reruns on the next batch.
Philosophy
Design principles.
Key steps
Brief to launch, in four steps.
Demo
The whole journey, in one walkthrough.
The GenStudio demo, from campaign brief to brand-applied assets, quality control, and final approval.
01. Land in GenStudio
The enterprise entry point: brand, campaigns, and credits in one place.
Launch
Adobe Summit, then the MAX keynote.
I demoed the suite to enterprise customers at Adobe Summit in Las Vegas, 2025. It featured in the CEO keynote, and personalization and the workflow builder shipped in enterprise GA.
Adobe Summit · Vegas 2025
CEO keynote
Personalization
Workflow builder
Close
Thank you.
Adobe Firefly Enterprise. GenAI creative production, end to end.
Shipped with
Sources
More work
Other projects worth a look.
