Designer: what to automate and which tools to use
Field: Marketing and content. Role hub. Checked: 2026-08-18. Tools change over time. Sources for this review are at the bottom of the page.
At a glance
Visual generation is getting cheaper very quickly. Taste, a broad visual reference library, solving the business problem, and client work become more valuable. The risks are an attractive image that misses the goal and unclear rights to AI-generated images.
What the data says (GDPval, Anthropic Economic Index, O*NET)
- Capability (GDPval and market evidence): image generation and routine work such as background removal, upscaling, and resizing are getting cheaper very quickly.
- Use (AEI): automation is growing in the basic production layer of visual work.
- The 2026 data argues with the simple "GenAI killed design" conclusion. The split runs inside the profession: employment growth for graphic designers is modest, around 2% through the mid-2030s, while UX roles are projected at about +16%. Design budgets are mostly holding or rising: 47% of businesses increased them, 32% kept them flat, and only 12% cut.
- The premium sits on design plus AI: designers with AI skills earn roughly 56% more than peers without them, and 69% of marketing and creative leaders say AI is reshaping the skills they need. What got cheaper is visual production, not design as a profession.
- Mode: augment moving toward replacement. Generic work loses value. Taste, art direction, a design system, and solving the business problem remain. See the full data review and the role AI strategy.
What to automate first
- Concepts and mood boards. Explore styles and directions quickly before detailed work.
- Image routine. Remove backgrounds, upscale, resize, and adapt assets for different formats.
- Variations and iterations. Generate many options quickly, then make the selection yourself.
- UI copy and icons. Draft interface text and generate icons from a description with Figma plugins.
- Social content visuals. Build templates and channel adaptations.
Task review: what AI can do and what you must check
| Task (O*NET) | Give to AI: method or tool | Keep or verify yourself | Prompt to start |
|---|---|---|---|
| Concepts and mood board | Generate directions | Select for the actual goal | Give me five directions for this idea. I will choose |
| Image routine | Background removal, upscaling, resizing | Result quality | |
| Variations | Produce many versions quickly | Taste and final choice | Give me five variations of this composition |
| UI copy and icons | Draft text and create icons from a description | Brand fit | Generate icons for these actions |
| Rights to AI images | Support the rights check | Decide whether commercial use is allowed |
Generation gets cheaper. Taste, art direction, and solving the business problem remain. Check the rights for every AI-generated image you plan to use.
Tools by use case
- ChatGPT or Claude for ideas, copy, and briefs.
- Midjourney, DALL-E, and Adobe Firefly for generation and editing; Figma AI and plugins such as Magician for UI copy and icons; Canva AI; and Looka for branding.
One practical playbook
From brief to concepts in one hour
- Write the brief, then generate five directions in Midjourney or Firefly for a mood board.
- Select one or two and refine the background, resolution, and composition in an editor.
- Human check: you own the fit with the business problem and the brand voice.
- Check whether the rights for each AI-generated image allow commercial use.
- Output: the research takes an hour. Taste and the decision remain yours.
Where not to use AI
Red flags
- Copyright and other rights for AI images: check the service terms and whether commercial use is allowed.
- Final taste and the design system: AI gives options. A person selects them.
- A client's brand guidelines: do not upload them to an open service without permission.
Prompt patterns: weak and better
| Weak | Better |
|---|---|
Make a logo |
Give me five directions for this idea. I will choose and refine one |
Make it look good |
Give me three solutions to this problem and explain the risk in each one |
Where to move your effort
Move time from production into art direction, the design system, the business problem, and defending the idea with the client.
The levels ladder in this role
The five levels from Chapter 5, in the language of this profession. Mark where you stand in your main tasks this week.
| Level | What it looks like here |
|---|---|
| 1. AI user | I generate visuals and clear the routine work: backgrounds, upscaling, resizing |
| 2. Validator | I select and finish. Generic output can be beautiful and still miss the brief; telling those apart is my job |
| 3. Orchestrator | I built the flow from brief to options to selection to finish, with AI supplying the volume of options |
| 4. Outcome owner | I own whether the design actually solves the business problem, not how many mockups I made |
| 5. System builder | I built the design system and rules that produce the right result without me |
Where people usually get stuck. Level 1 is where people stop: plenty of options, no taste being trained. Art direction and defending an idea to the client are what gain value.
Next: my level of usefulness → a plan for one level up.
Ready-made skills and plugins for this role
You can turn repeatable procedures such as mood boards and concepts, asset resizing and adaptation, and brand-guideline checks into a portable skill, or use an existing one. Browse skill banks, then see turn a workflow into a skill to build your own.
Where to go next
Map your week → run the integrated profession audit → make a 90-day plan. Or open the workbook.
Sources for this review: OpenAI GDPval (2025): https://openai.com/index/gdpval/ · Anthropic Economic Index: https://www.anthropic.com/economic-index · Stanford AI Index (2025): https://hai.stanford.edu/ai-index · McKinsey, "The State of AI": https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai. Plus the book, Chapters 2, 3, and 7. Version: 2026-08-18.