Community Field Guide
Tools We Use
A practical, opinionated guide to the AI tools shaping how designers research, create, prototype, and build.
Tested by people doing the work. Updated as the tools and our opinions change.
Why This List Exists
The useful tools earn their place
Not everything is worth your attention
New AI tools appear every week. Most do not deserve a permanent place in a designer's workflow. This is a deliberately selective list of products that members of the AI Design LA community have used enough to form an honest point of view about.
How We Classify Them
What the labels mean
Tools people regularly depend on. Tested across enough real projects to trust.
Useful for particular workflows, projects, or moments — but not every one.
Promising, but not yet proven enough for regular use. Worth keeping an eye on.
Tools that were replaced, declined, or created more friction than value.
Currently in the mix
The toolkit
Browse the tools our community is using across the design process.
Think & Research
Anthropic's conversational AI, used across the community for research synthesis, writing, strategic thinking, and working through design problems in plain language.
Best for
Deep reasoning about design decisions, synthesizing research, writing briefs and specs, and thinking through complex product problems before touching a tool.
Community take
The model people reach for when the problem is genuinely hard and needs thinking, not just generation. Better at 'why' than almost anything else.
A research repository and analysis platform with AI-assisted tagging, summarization, and insight extraction from interviews and notes.
Best for
Teams running regular research who want to build a shared, searchable repository of insights over time.
Community take
The AI features are genuinely useful but not transformative yet. The real value is the repository habit it enforces, not the AI layer on top.
Google's flagship AI model with deep integration across Workspace products and strong multimodal reasoning capabilities.
Best for
Designers embedded in the Google ecosystem, working with documents and sheets, or needing strong web search integration in reasoning tasks.
Community take
Most useful inside Workspace rather than as a standalone. A reasonable alternative when the problem involves Google-specific tools or data.
An AI-powered search and research tool that cites its sources, making it more trustworthy for desk research than general-purpose models used in browse mode.
Best for
Competitive analysis, technology landscape research, and any task where you need a quick, sourced answer rather than a thorough model response.
Community take
More useful than a search engine for research phases. Less useful for reasoning. The citation habit is its biggest differentiator.
Design & Prototype
A prompt-driven environment inside Figma for creating interactive concepts and exploring product ideas quickly without leaving the design tool.
Best for
Early interaction design, rapid concept testing, and communicating behavior before committing to production code.
Community take
Fast and surprisingly capable. Strong design judgment is still required — it accelerates the work but doesn't replace the thinking.
Framer
Visit site ↗A design and publishing tool with AI-assisted layout generation, component creation, and one-click site deployment.
Best for
Marketing sites, portfolio work, and interactive prototypes that need to be shareable and real — not just clickable mocks.
Community take
The gap between prototype and shipped site is genuinely smaller here. The AI features are useful for layout, less so for complex interaction design.
Paper
An emerging interface design environment focused on rapid visual exploration and a more fluid canvas model than established design tools.
Best for
Designers exploring alternatives to Figma who want a more generative approach to visual exploration.
Community take
Visually promising. Still needs broader testing on real product work before it earns a place in a production workflow.
Build
An agentic coding tool that helps designers move from design intent into working software, modify real codebases, connect services, and build prototypes that can grow beyond a disposable demo.
Best for
Designers who want to build directly in code or collaborate more deeply with engineers on real production work.
Community take
Extremely capable, but not magic. Works best when the designer provides clear requirements, real components, and disciplined feedback. The quality of your input determines the quality of the output.
Cursor
Visit site ↗An AI-powered code editor for understanding, generating, and modifying code within an existing project, with deep codebase context awareness.
Best for
Making targeted interface changes, navigating unfamiliar codebases, and iterating with more control than a one-shot app builder.
Community take
A strong bridge into code for designers. Some technical comfort is still necessary, but less than a traditional editor requires.
GitHub Copilot
Visit site ↗An AI pair programmer embedded in most major code editors, offering inline completions and chat-based coding assistance.
Best for
Designers who are already writing code and want faster autocomplete and inline suggestions without switching tools.
Community take
Most useful when you already know what you want to write. Less useful when you are exploring or problem-solving — use Claude for that.
A prompt-based web app builder for creating functional MVPs and testing product ideas quickly without writing code from scratch.
Best for
Early product validation, lightweight internal tools, and prototypes that need real data or interactions to be convincing.
Community take
Impressive speed at the beginning. Control, reliability, and maintenance become harder as complexity grows. Know when to move to a real codebase.
Generate
Adobe Firefly
Visit site ↗A suite of generative image and editing capabilities integrated across Adobe's creative products, trained on licensed imagery.
Best for
Working with existing creative assets, production editing, compositing, and commercially oriented workflows where rights matter.
Community take
Most valuable inside Photoshop and existing creative workflows rather than as a standalone image generator. The licensing story is its real differentiator.
Midjourney
Visit site ↗A generative image tool with strong aesthetic range and a distinctive visual sensibility, used for concept exploration and moodboarding.
Best for
Early-stage concept imagery, moodboards, art direction exploration, and generating reference material that informs rather than ships.
Community take
Still the best at producing images that have a point of view. The challenge is avoiding the aesthetic becoming a crutch rather than a starting point.
Runway
Visit site ↗A creative AI platform for generating and editing video, with motion generation and video-to-video transformation tools.
Best for
Concept videos, animatics, and motion explorations where production-quality video is premature but static images are not enough.
Community take
The technology is genuinely impressive. Integrating it into a real design process is still a work in progress for most people.
Test & Evaluate
Maze
Visit site ↗A rapid usability testing platform with AI-assisted analysis, path analysis, and unmoderated study capabilities.
Best for
Teams who need to run frequent, lightweight usability tests and want faster synthesis than manual review allows.
Community take
The AI analysis is a genuine time-saver for large studies. For small or nuanced studies, it can flatten findings that deserve more careful reading.
The Tool Graveyard
What we tried and left behind
Some tools are replaced. Some stop improving. Others create more work than they save. These are the ones we stopped reaching for and why.
Diagram (Magician for Figma)
Replaced by
Native Figma AI features
Former use
AI-assisted design exploration inside Figma
Its most valuable capabilities were gradually absorbed into Figma's own AI features. Maintaining a separate plugin stopped making sense.
Early AI writing assistants
Replaced by
Claude and GPT-4o
Former use
UX copy and microcopy generation
Generic outputs that required more editing than writing from scratch. The model improvement curve shifted this — current models are meaningfully better.
Early one-shot app builders
Replaced by
Claude Code and Cursor
Former use
Instant product demos and concept prototypes
They produced impressive first results but became difficult to control, extend, or maintain as complexity grew. The demo was not the product.
Stable Diffusion (local)
Replaced by
Midjourney and Adobe Firefly
Former use
Image generation with full prompt control
The setup and maintenance overhead was not worth the control advantage for most design use cases. Cloud tools caught up faster than expected.
Uizard
Replaced by
Figma Make and Lovable
Former use
Rapid interface generation from sketches and screenshots
The output felt generic and required significant cleanup before it was useful in a real project. The ratio of editing to generating tilted the wrong way.