Take an idea or prompt and get working screens, flows, or front-end code. Each tool sits at a different point on the "vibe to production" spectrum.
Figma Make
Prompt-driven prototype and mini-app generator built inside Figma, with Make Kits and Make Attachments pulling in real components and context.
Pros
Lives inside the file you already work in — no tool-switch tax.
Make Kits let you start from your real design system rather than generic output.
Agent + MCP support means AI can write back to the canvas, not just suggest.
Cons
Output quality still drops off for anything more than landing-page complexity.
Plan tiering for Make is confusing; seats and usage caps hit fast on teams.
Best for: designers who want AI inside Figma, not a separate tabPricing: Included in Figma paid plans; Make quotas varyfigma.com/ai
v0 by Vercel (v0.app)
Chat-driven React and Next.js component generator. Rebranded from v0.dev to v0.app in January 2026 as a full-stack development surface.
Pros
Output is production-quality React — components integrate with real codebases.
Strong at Tailwind + shadcn/ui patterns used across modern front-ends.
Tight loop with Vercel deploys and Next.js conventions.
Cons
Assumes you know React — not a no-code tool despite the chat interface.
Opinionated on stack; swimming against Tailwind or shadcn is painful.
Best for: designers who code, or designers handing off to React teamsPricing: Free tier + paid from $20/mov0.app
Lovable
Prompt your way to a deployed, full-stack web app with a backend, auth, and database. Reached $20M ARR in two months — fastest growth in European startup history.
Pros
Goes from idea to a live, working product without writing code.
Includes backend, auth, and database out of the box (Supabase integration).
Great for internal tools, MVPs, and founder-mode prototypes.
Cons
Output is harder to own and migrate off than v0's plain React.
Design polish is generic — you'll need to push styling to match brand.
Best for: non-technical builders shipping working productsPricing: Free starter + paid from $20/molovable.dev
Bolt.new
Fast prompt-to-app builder from StackBlitz, with particular strength in mobile-first prototypes.
Pros
Fastest first result among the "prompt an app" tools.
Solid mobile and responsive output.
In-browser dev environment — no local setup.
Cons
Less opinionated than v0 or Lovable — you'll need to steer the architecture.
Iteration on complex apps can get brittle over long sessions.
Best for: rapid prototypes, especially mobilePricing: Free tier + paid from $20/mobolt.new
Google Stitch (v2)
Google's AI-native infinite canvas with voice input. The closest tool to the "describe a vibe, get a direction" promise.
Pros
Free — great for exploratory work before tool commitment.
Voice canvas feels native, not bolted on.
Genuinely good at early-stage visual exploration when you don't know what you want.
Cons
Not built for production-quality output or code.
Handoff to Figma or code is still manual.
Best for: 20-minute "vibe passes" at the start of any projectPricing: Freestitch.withgoogle.com
Claude Design (Anthropic)
Visual design surface from Anthropic, launched April 17 2026 on top of Claude Opus 4.7. Generates prototypes, slide decks, and marketing assets from prompts.
Pros
Tightly coupled to Claude's reasoning — good at following detailed briefs.
Useful for quick stakeholder-facing artefacts (decks, one-pagers) that normally steal design hours.
Backed by the 1M-token context window — you can feed it a full brand book.
Cons
Brand-new; features and export paths are still rough.
Positioned at non-designers — serious designers may find the primitives thin.
Best for: PMs and founders making design without designersPricing: Included with Claude paid plansclaude.ai
Framer AI
Prompt-to-website builder with strong animation and interaction primitives baked in.
Pros
Best-in-class for marketing sites with polished motion.
Direct publish to a real domain, no export step.
CMS and localisation built in for content-heavy pages.
Cons
Not ideal for apps — strength is in websites.
Lock-in risk: exporting to plain code is not Framer's strong suit.
Best for: marketing sites, portfolios, landing pagesPricing: Free + paid from $5/mo (site) / $20/mo (pro)framer.com
Magic Patterns
Generates React + Tailwind UI patterns from prompts with strong component-library coverage.
Pros
Good at generating individual components rather than whole apps.
Integrates with popular libraries (shadcn, MUI, Chakra).
Export to code is cleaner than most.
Cons
Smaller team, less polish than v0 or Lovable.
Best for component-sized work, not full flows.
Best for: generating individual UI patterns during designPricing: Free tier + paid from $20/momagicpatterns.com
Design Systems
Tools that generate, audit, or extend a design system rather than individual screens. The frontier here is generative tokens and auto-documentation.
Relume
AI-generated sitemaps and wireframes that produce ready-to-use components for Webflow, Figma, and Framer.
Pros
Huge library of pre-built, accessible components.
Goes from brief to full site structure in minutes.
Excellent for agency work and client-facing IA proposals.
Cons
Aesthetic converges on a "Relume look" if you don't override it.
Subscription is on the pricier side for individual designers.
Best for: agencies pitching site structures fastPricing: Paid from $30/morelume.io
Subframe
Component design tool that outputs production-ready React and Tailwind, positioned as "Figma for engineers."
Pros
Output is clean React code — not screenshots or traced markup.
Tight visual-to-code parity; no design-dev drift.
Strong prop and variant model.
Cons
Learning curve if you're coming from pure Figma.
Smaller ecosystem; fewer plugins and community templates.
Best for: design-system teams shipping React componentsPricing: Free tier + paid from $19/mosubframe.com
Figma AI Design Systems Generator
Figma Make's design-system-aware mode — produces tokens, components, and documentation starters from a brand brief.
Pros
Generates tokens and components that follow your brand constraints.
Saves days of naming and documentation grunt work.
Sits inside Figma where your team already works.
Cons
Works best starting from scratch — retrofitting an existing system is clunky.
Output still needs a human pass for tokenisation discipline.
AI-native design tool marketed as a Figma alternative, with emphasis on smart auto-layout and component suggestions.
Pros
AI suggestions for spacing, alignment, and component usage are genuinely useful.
Free for personal use, lower ceiling for teams.
Fast rendering on large files.
Cons
Ecosystem is still small — fewer plugins, less community material.
Moving an established team off Figma is a real cost.
Best for: small teams willing to try a Figma challengerPricing: Free + paid from $8/mo/editormotiff.com
MCP Servers
Model Context Protocol servers expose tool-specific data and actions to AI assistants. A good MCP is the difference between an AI that guesses and one that works with your actual files and systems.
Figma MCP (official)
Exposes live Figma selection structure — hierarchy, auto-layout, variants, text styles, spacing tokens, component references — to any MCP-compatible AI.
Pros
Turns design handoff from screenshot-guessing into structured code generation.
AI can read and now write back to files through agent mode.
Maintained by Figma — updates track platform changes.
Cons
Requires Dev Mode seats for full capability.
Performance on very large files still drops.
Best for: any design-to-code workflowPricing: Included in Figma Dev ModeFigma MCP Guide
AIDesigner MCP
Community MCP server focused on AI-powered UI generation tasks, often paired with Claude Code and Cursor.
Pros
Purpose-built for design-adjacent prompts and tasks.
Pairs well with a coding agent to turn prompts into working UI.
Cons
Community-maintained — reliability varies with upstream changes.
Less useful if you're already fluent with Figma MCP.
Best for: designers prototyping inside a coding agentPricing: Free (self-hosted)aidesigner.ai
GitHub MCP
Lets an AI read PRs, issues, commits, and code across your repos — the connective tissue for any design-to-code workflow.
Pros
Gives the AI real context on implementation, not just design intent.
Maintained by GitHub — well-behaved auth and rate limits.
Essential for any designer collaborating with engineers via code.
Cons
Overlaps with built-in tooling in Cursor/Windsurf — sometimes redundant.
Permissions sprawl if you give the agent write access.
Integrated into Gemini and Firefly's model picker.
Cons
Quality ceiling is lower than Midjourney or Flux for hero imagery.
Priced higher than expected for bulk use (~$0.08/image).
Best for: speed-sensitive workflows, in-product image featuresPricing: ~$0.08/image via APIdeepmind.google
Video Generation
The field moved fast: native 4K, synchronised audio, and multi-shot storyboards are table stakes in 2026. OpenAI announced Sora's web and app would shut down — Veo and Kling are eating the market.
Sora 2 (OpenAI)
OpenAI's cinematic-quality video model with realistic physics and synchronised audio. Note: OpenAI announced in April 2026 it will discontinue the Sora web and app experiences, with API access ending September 2026.
Pros
Still unmatched for narrative coherence and realism.
Prompt adherence is exceptional for complex scenes.
Cons
Sunsetting — not a tool to build a new workflow around.
Access is already being throttled as the shutdown approaches.
Best for: one-off use only while it's still aroundPricing: Included with ChatGPT Pro (access limited)openai.com/sora
Google Veo 3.1
Native 4K (3840×2160) video at up to 60fps, with synchronised audio and strong character consistency across longer clips (up to 30 seconds).
Pros
Highest-resolution output on the market.
Character consistency is the best of any current model.
Audio is generated in sync, not bolted on.
Cons
Access gated through Google — workflows can be clunky.
Pricey for bulk commercial use.
Best for: product launch videos, brand films, any 4K deliverablePricing: Via Gemini paid tiers / Vertex AIdeepmind.google/technologies/veo
Kling 3.0
Released February 4, 2026. Headline feature is Multi-Shot Storyboard — define a whole sequence with individual prompts and camera angles in one pass.
Pros
Best value-for-money in AI video — Standard plan starts at $6.99/mo.
Multi-shot storyboard is a unique, genuinely useful feature.
High-volume social output without breaking the budget.
Cons
Resolution tops out below Veo's native 4K.
Fewer camera-control primitives than Runway.
Best for: social content, storyboarded sequences, volume workPricing: From $6.99/mokling.kuaishou.com
Runway Gen-4.5
The filmmaker's pick. Prioritises direct creator control over camera moves, lighting shifts, and scene transitions.
Pros
Best granular control over motion, camera, and transitions.
Deep editing suite alongside the generator.
Trusted by film and ad agencies.
Cons
Pricing gets steep at production volume.
Learning curve — not a prompt-and-go experience.
Best for: filmmakers, VFX, directed video workPricing: Paid from $15/mo, enterprise tiers aboverunwayml.com
3D Generation
No single winner — tools specialise by output type. Pick by whether you need game-ready topology, photoreal visualisation, or interactive scenes for the web.
Meshy
Fast text- and image-to-3D with a solid editing workflow. Strong for iteration.
Pros
Fastest generation loop for iterating on shapes.
Best features-per-dollar in the category.
Commercial rights on paid plans.
Cons
Topology is OK but not always game-ready.
Photorealism lags behind Rodin.
Best for: concept 3D, AR, fast iterationPricing: Free tier + Pro from $14.50/momeshy.ai
Tripo
Generates clean quad-based topology — the format games and real-time engines actually want.
Pros
Game-ready topology out of the box.
Cheapest paid option at $12/mo.
Works well with Blender, Unity, Unreal pipelines.
Cons
Aesthetic ceiling is lower than Rodin for hero renders.
Smaller community; fewer tutorials than Meshy.
Best for: game assets, real-time 3D, AR/VRPricing: From $12/motripo3d.ai
Rodin
Top of the category for photorealistic object generation. Aimed at professional visualisation work.
Pros
Best photoreal output in AI 3D.
Strong material and texture fidelity.
Cons
Expensive ($120/mo Business tier) — only justifies for pro visualisation.
Overkill for concept work or games.
Best for: product visualisation, e-commerce hero rendersPricing: Business from $120/mohyperhuman.deemos.com
Luma AI
Strong at scene captures from video and photos. Generous free tier makes it a great entry point.
Pros
Best in class for capturing real-world scenes as 3D.
Generous free tier for learning.
Great for AR and spatial content.
Cons
Generated (vs captured) 3D is not its strength.
Output resolution varies with capture quality.
Best for: capturing real objects and scenes, AR contentPricing: Free + paid tierslumalabs.ai
Spline AI
Real-time 3D editor with collaboration, physics, and interactive behaviours — built for the web.
Pros
Real-time collaboration like Figma, but for 3D.
Export and embed straight into web projects.
AI features for generating materials and simple scenes.
Cons
Not a generation-first tool — more of a 3D design tool with AI features.
Can't match dedicated generators for scene fidelity.
Best for: interactive 3D on websites, onboarding animationsPricing: Free tier + paid from $9/mospline.design
Coding Agents
Designers increasingly ship code. These agents are how you go from Figma to a working implementation — or audit existing code against a design.
Cursor
AI-first IDE. The default choice for designer-developers in 2026.
Pros
Best-in-class editor experience with deep AI integration.
Works with Figma MCP and other design MCPs out of the box.
Strong at long multi-file refactors.
Cons
Subscription stacks on top of API costs if you use premium models.
Not a design tool — no canvas.
Best for: designers shipping real codePricing: Free tier + Pro $20/mocursor.com
Claude Code
Anthropic's CLI-native coding agent. Runs in the terminal and works on real projects.
Pros
Built on Opus 4.7 with the 1M-token window — can hold a whole codebase in mind.
Excellent at autonomous multi-step tasks.
Unix-philosophy fit — composes with shell tools and MCPs.
Cons
Terminal-first UX is a step up for many designers.
Can burn tokens fast on exploratory sessions.
Best for: experienced designers comfortable in a terminalPricing: Included with Claude Pro or pay-per-use via APIclaude.com/product/claude-code
Replit Agent
Browser-based coding agent that builds and deploys apps without leaving the tab.
Pros
Zero local setup — works anywhere there's a browser.
Includes hosting and deploy out of the box.
Great for PMs and designers who want to ship without a dev env.
Cons
Less suitable for complex or long-lived codebases.
Egress to your own infra can be painful later.
Best for: quick prototypes, internal tools, zero-setup workPricing: Free tier + paid from $20/moreplit.com
Windsurf (Codeium)
Cursor's main competitor — AI-native IDE with strong autonomous mode ("Cascade").
The notes actually reflect what mattered, not just a transcript.
Native Mac app; no bot joining the call.
Searchable meeting history becomes a real resource.
Cons
Mac-only currently.
Privacy posture varies by company; check with legal.
Best for: PMs and designers drowning in meetingsPricing: Free tier + paid from $18/mogranola.ai
UX Research
88% of UX researchers list AI-assisted analysis as a top trend for 2026. The split is between real-user research platforms adding AI features and AI-native platforms launched post-ChatGPT.
Dovetail AI
Research repository with AI-powered synthesis across interviews, transcripts, and notes.
Pros
Mature repository — your research becomes a living asset.
Synthesis and theme detection save days per project.
Good export paths for stakeholder reports.
Cons
Priced for teams, not individual researchers.
AI synthesis can paper over tension that matters — always review.
Best for: research teams with ongoing studiesPricing: Paid from $30/user/modovetail.com
Maze AI
Usability testing platform with AI-moderated interviews and auto-generated test reports.
Pros
AI moderator conducts unmoderated-style sessions that still adapt to responses.
Fast turnaround from test to report.
Integrations with Figma prototypes.
Cons
AI moderation misses the subtle probing a human researcher would do.
Quant leans are real — qual depth is shallower than a human-led study.
Best for: fast directional usability testingPricing: Free tier + paid from $99/momaze.co
Synthetic Users
Runs research with AI-simulated users. Lets you explore directional insights before committing to real recruitment.
Pros
Useful for early exploration when real recruits aren't worth the cost.
Helps stress-test interview guides before running them live.
Fast — directional findings in hours, not weeks.
Cons
Not a substitute for real users — known failure modes around surprise and nuance.
Easy to mistake confident-sounding output for truth.
Best for: directional validation, interview-guide testingPricing: Paid from $75/mosyntheticusers.com
UserTesting AI Insights
UserTesting's AI layer over its real-user panel — auto-generates highlights, themes, and clips from sessions.
Pros
Real users, real qualitative data — AI layer accelerates synthesis.
Very large panel for targeted recruiting.
Good at generating shareable clips for stakeholders.
Cons
Expensive — enterprise pricing.
Auto-themes sometimes miss the actual problem; researcher review needed.
Best for: large organisations with formal research practicesPricing: Enterprise; contact salesusertesting.com