Design-Generator Catalog

The AI app/website design generators a design brief can target. Use this to pick the right tool for a brief's constraints, and as the per-tool input spec the brief-tailoring layer drives from. Facts below reflect vendor docs as of the last-reviewed date; this market churns fast, so treat anything older than the refresh cadence as suspect.

Quick chooser

Tool	Generates	Stack choice	Backend logic	Design-system fidelity	Best fit
v0	Full-stack web apps, deployable to Vercel	Next.js/React/Tailwind/shadcn (fixed)	Yes — integration-based (Supabase et al.)	High (coded) — shadcn registries + design tokens	Next.js teams wanting on-brand, production-adjacent UI
Lovable	Full-stack web apps with built-in cloud	React/Vite/Tailwind/Supabase (fixed)	Strongest — native Postgres/auth/storage/edge functions	Low–moderate — prompt/Knowledge steering only	Non-developers shipping real MVPs (auth, DB, payments)
Bolt	Web apps in your chosen JS framework, plus Expo mobile	Widest — React, Vue, Next.js, Astro, Svelte, Remix, Expo	Yes — Bolt Cloud built-ins + Supabase	Moderate — conventions by prompt/Knowledge, no token ingestion	Developers wanting stack choice or mobile output
Figma Make	React/TypeScript prototypes and light apps	React/TS (fixed)	Limited — Supabase-backed, prototype-grade	High (visual) — imports Figma library variables/styles	Designers validating on-brand interactive prototypes

Static/frontend-only briefs suit any of the four; v0 and Figma Make give the strongest visual results. Backend-logic briefs (DB, auth, real workflows) suit Lovable or Bolt first, v0 where the team is already on Vercel, and Figma Make only for prototype-grade validation.

v0 (Vercel)

Vercel's agentic app builder at v0.app (renamed from v0.dev in 2025; no longer just a shadcn component generator).

Aspect	Detail
Input	Plain-prose prompts; screenshots/images, Figma files, pasted code as attachments; project-level instructions
Design-system input	shadcn registries — a documented spec for feeding components, blocks, and design tokens (Tailwind config + CSS variables); the strongest coded-token story of the four
Output	Next.js + React + Tailwind + shadcn/ui full-stack apps; Git/PR workflow and direct Vercel deploy
Pricing shape	Subscription tiers with token-metered credits

• Strengths: idiomatic, production-adjacent code in its home ecosystem; real design-token conformance via registries; agentic (web search, error fixing, planning); tight deploy/Git loop.
• Limitations: vendor docs warn it is trained on stock shadcn components and may struggle with customizations; output is Next.js/Vercel-shaped — other stacks are second-class; backend wiring is integration-based, thinner than Lovable/Bolt's built-in clouds.
• Backend vs. static: capable full-stack, but its center of gravity is frontend; best when the backend is Vercel-ecosystem integrations.
• Best fit: teams on Next.js/Vercel/shadcn wanting a design-system-aware component factory or prototype-to-PR flow.

Lovable

Full-stack "prompt to working app" platform (lovable.dev) with its own Supabase-backed cloud.

Aspect	Detail
Input	Plain prose + images; Chat mode (plan, no code changes) vs build mode; per-element visual edits; a project Knowledge file for persistent context; Figma import via the Builder.io plugin
Prompting conventions	Vendor teaches component-first prompts (section-by-section beats whole-page), atomic UI language, real content over lorem ipsum, plan-in-chat-mode-first
Output	React + Vite + Tailwind frontend; Lovable Cloud backend (Postgres, auth, storage, edge functions) or bring-your-own Supabase; two-way GitHub sync
Pricing shape	Subscription with credits; cloud usage billed separately — cost predictability is the top user complaint

• Strengths: fastest idea→working-full-stack-app path of the four — auth, DB schema, and deploy genuinely from one prompt; agent mode; full code ownership via GitHub sync.
• Limitations: locked to its React/Vite/Tailwind/Supabase stack; no design-token ingestion — brand fidelity rides on prompts, the Knowledge file, or imported Figma structure; credit burn is hard to predict on complex iterations.
• Backend vs. static: the strongest backend story of the four for non-developers — native Postgres, auth, RLS, storage, edge functions, payments connectors.
• Best fit: founders/PMs/small teams shipping real MVPs and SaaS apps without infra work.

Bolt (bolt.new)

StackBlitz's in-browser AI builder running on WebContainers (a real Node.js dev environment in the browser); Bolt Cloud provides built-in hosting/DB/auth.

Aspect	Detail
Input	Plain prose; built-in Enhance prompt refinement; Discussion mode to plan without burning build tokens; image attachments; Figma import; project/team Knowledge settings for persistent rules
Prompting conventions	Vendor guidance: name the stack up front ("Astro, Tailwind, shadcn" in the first prompt), one change per prompt, be explicit about what must not change, lock files you don't want touched
Output	Any WebContainers-capable JS framework — React, Vue, Next.js, Astro, Svelte, Remix — plus native mobile via Expo; deploy via Bolt Cloud or Netlify; GitHub sync
Pricing shape	Token-metered (generous free tier); token cost scales with project size — the top user complaint on large projects

• Strengths: the only one of the four with real stack choice, including mobile; a live dev environment that installs packages and runs the app while generating; built-in DB/auth/storage via Bolt Cloud plus first-class Supabase.
• Limitations: confined to the Node/JS universe (no Python/Go/JVM backends); token burn grows with project size, so long sessions need context hygiene; UI polish depends heavily on prompt specificity — historically weaker default "taste" than v0.
• Backend vs. static: genuine full-stack within Node/JS limits.
• Best fit: developers and technical tinkerers who want to pick the framework, need Expo mobile output, or want a live runnable environment.

Figma Make

Figma's prompt-to-app tool, generally available since mid-2025; aimed at designers, bundled with paid Figma seats.

Aspect	Detail
Input	Plain-prose chat; pasted Figma frames/components (the headline input mode); imported variables and styles from Figma Design libraries; image paste; URL fetch for real data
Editing	Point-and-click tweaks on the generated canvas (type, color) without prompting
Output	React/TypeScript web apps and functional prototypes; publishable to a hosted URL; code viewable in-app and downloadable as a zip — no Git/PR workflow
Pricing shape	Included with paid Figma seats (monthly AI credits); Full seat required to create/publish

• Strengths: lives where designers work — pasted frames plus library variables/styles give it design context no competitor gets natively; zero-handoff interactive prototypes with working logic; Supabase integration adds real auth/DB/file storage.
• Limitations: generated code is criticized as non-component-structured ("div soup") needing rewrites for production; weakest engineering loop of the four (zip export, no Git); modest monthly AI credits for heavy iteration.
• Backend vs. static: limited but real via Supabase — prototype-grade, not positioned for production backends.
• Design-system fidelity: strong on visual fidelity (library variables, styles, components keep output on-brand) but weak on code-level fidelity — output does not map to a coded component library or token pipeline.
• Best fit: design/PM teams inside Figma validating high-fidelity, on-brand interactive prototypes without engineers.

Choosing from brief constraints

How a brief's constraints section maps to candidates:

Brief constraint says…	Suggest
Next.js / Vercel / shadcn stack, coded design tokens exist	v0
Needs real backend (auth, DB, payments), non-developer driver	Lovable
Specific non-Next framework (Vue, Astro, Svelte…) or mobile (Expo)	Bolt
Figma design system exists; goal is a validated prototype, not shipped code	Figma Make
Static marketing/landing page, visual polish paramount	v0 or Figma Make
Strict WCAG target	Any — none enforces accessibility; keep the requirement explicit in the brief and audit the output

Maintenance

• Owner: the owner in this doc's front matter.
• Staleness signal: the front-matter last_reviewed date — every claim here was checked against vendor docs on that date.
• Refresh cadence: review quarterly. This market reprices, renames, and re-platforms in months (v0.dev→v0.app, Lovable Cloud, Bolt v2, Figma Make GA all landed within one year); at review, re-verify each entry against the vendor links below and restamp last_reviewed via timelord.
• Off-cycle triggers: a brief-tailoring failure traced to a stale input convention, a vendor rename/GA/pricing-model change, or adding a new generator to the catalog.

Vendor documentation

• v0: v0.app/docs · design-systems guide
• Lovable: docs.lovable.dev · prompting guide
• Bolt: support.bolt.new — prompting effectively
• Figma Make: figma.com/blog — GA announcement · help center