AI Tools Comparison

A working shortlist of AI tools for designers and PMs, with honest pros and cons. Organised by category; last updated April 19, 2026.

Jump to category

Prototyping & UI Generation

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 tab Pricing: Included in Figma paid plans; Make quotas vary figma.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 teams Pricing: Free tier + paid from $20/mo v0.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 products Pricing: Free starter + paid from $20/mo lovable.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 mobile Pricing: Free tier + paid from $20/mo bolt.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 project Pricing: Free stitch.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 designers Pricing: Included with Claude paid plans claude.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 pages Pricing: 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 design Pricing: Free tier + paid from $20/mo magicpatterns.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 fast Pricing: Paid from $30/mo relume.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 components Pricing: Free tier + paid from $19/mo subframe.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.
Best for: new design systems or system audits Pricing: Included in Figma paid plans figma.com/solutions/ai-design-systems-generator

Motiff

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 challenger Pricing: Free + paid from $8/mo/editor motiff.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 workflow Pricing: Included in Figma Dev Mode Figma 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 agent Pricing: 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.
Best for: designers reviewing implementation against designs Pricing: Free github.com/github/github-mcp-server

Playwright MCP

Browser automation exposed to AI. Lets an agent open a page, take screenshots, inspect DOM, and click through flows.

Pros

  • Unlocks visual QA and accessibility audits from a chat prompt.
  • Works with any web property — not tool-specific.
  • Pairs beautifully with design-review workflows.

Cons

  • Setup is more involved than most MCPs — Playwright dependencies to manage.
  • Slow compared to DOM-API tooling for quick checks.
Best for: visual QA, accessibility audits, design review Pricing: Free (open source) github.com/microsoft/playwright-mcp

Supabase MCP

Gives an AI read/write access to a Supabase project — rows, schema, migrations, edge functions.

Pros

  • Designer can ask real-data questions ("how many users actually use this feature?") without a PM or analyst.
  • Great for data-driven design reviews.
  • First-party, maintained.

Cons

  • Assumes your backend is Supabase.
  • Write access is risky in shared projects — scope carefully.
Best for: designers on Supabase-backed products Pricing: Free (Supabase plan separate) supabase.com/docs

Image Generation

Specialised tools have split the space: pick by the job (aesthetic direction, brand, text-in-image, vectors, speed).

Midjourney V8

Still the reference point for aesthetic art direction. V8 added native 2K output and is 5x faster than V6.

Pros

  • Best for "wow" output that doesn't look generic.
  • Strong style references and sref system for brand consistency.
  • Large prompt vocabulary built up over years of use.

Cons

  • Text rendering is still its weak spot.
  • Discord-first UX is clunky for production workflows.
Best for: moodboards, hero imagery, editorial Pricing: Paid from $10/mo midjourney.com

Flux 2 (Black Forest Labs)

32-billion-parameter model focused on photorealism — depth of field, lens distortion, chromatic aberration, film grain.

Pros

  • Best-in-class skin textures and realistic lighting.
  • Strong character consistency across variations.
  • Available via multiple API providers at reasonable cost (~$0.03/image).

Cons

  • Less controllable than Midjourney for stylised art.
  • Open-weight provenance creates brand caution in some orgs.
Best for: photorealistic stock, character scenes, product shots Pricing: ~$0.03/image via API providers blackforestlabs.ai

GPT Image 1.5 (OpenAI)

OpenAI's image model with strong prompt adherence and the best text-in-image rendering on the market.

Pros

  • Best at following detailed, multi-constraint prompts.
  • Text rendering is reliable enough for production brand work.
  • Available inside ChatGPT and via API.

Cons

  • Aesthetic ceiling is lower than Midjourney's.
  • Content moderation is stricter — some legitimate prompts get flagged.
Best for: complex prompts, text in images, brand-accurate output Pricing: Included with ChatGPT Plus, API priced per image platform.openai.com

Recraft V4

The best AI model for logos and vector work in 2026. Exports SVG and has built-in brand styling.

Pros

  • Native SVG export — scalable, editable, production-ready.
  • Top of HuggingFace benchmarks for logo generation.
  • Brand style memory lets you keep consistent looks across assets.

Cons

  • Narrower use case — not a general-purpose image tool.
  • Still produces "AI logo" tells on close inspection.
Best for: logo exploration, icon systems, vector illustration Pricing: Free tier + paid from $12/mo recraft.ai

Ideogram 3.0

Specialised in text inside images, with 90–95% typography rendering accuracy. Strong for posters, social graphics, branded content.

Pros

  • Near-perfect typography inside generated images.
  • Good for social content, quote cards, event posters.
  • Reasonable pricing, friendly UI.

Cons

  • Aesthetic range narrower than Midjourney or Flux.
  • Less powerful for purely photographic output.
Best for: posters, social graphics, anything with text baked in Pricing: Free tier + paid from $7/mo ideogram.ai

Nano Banana 2 (Gemini 3.1 Flash Image)

Google's fastest image model — ~1–3 seconds per image. Part of the Gemini family.

Pros

  • Fastest generation on the market — useful for iteration and UI embeds.
  • Strong editing capabilities (inpainting, object removal).
  • 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 features Pricing: ~$0.08/image via API deepmind.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 around Pricing: 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 deliverable Pricing: Via Gemini paid tiers / Vertex AI deepmind.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 work Pricing: From $6.99/mo kling.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 work Pricing: Paid from $15/mo, enterprise tiers above runwayml.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 iteration Pricing: Free tier + Pro from $14.50/mo meshy.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/VR Pricing: From $12/mo tripo3d.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 renders Pricing: Business from $120/mo hyperhuman.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 content Pricing: Free + paid tiers lumalabs.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 animations Pricing: Free tier + paid from $9/mo spline.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 code Pricing: Free tier + Pro $20/mo cursor.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 terminal Pricing: Included with Claude Pro or pay-per-use via API claude.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 work Pricing: Free tier + paid from $20/mo replit.com

Windsurf (Codeium)

Cursor's main competitor — AI-native IDE with strong autonomous mode ("Cascade").

Pros

  • Cascade mode handles multi-file agentic tasks well.
  • Competitive pricing and often better enterprise controls.
  • Good inline completion quality.

Cons

  • Smaller community than Cursor; fewer third-party integrations.
  • Changing IDEs is a real switching cost.
Best for: teams with stricter compliance or enterprise needs Pricing: Free tier + paid from $15/mo windsurf.com

PM & Product Tools

If you partner with PMs, the artefacts they hand you are changing. These tools generate PRDs, synthesise feedback, and manage roadmaps with AI.

ChatPRD

AI assistant specialised for writing, reviewing, and iterating on product specs and PRDs.

Pros

  • Tuned specifically for PM workflows — better structure than generic LLMs.
  • Generates acceptance criteria, user stories, and success metrics.
  • Integrates with common PM stacks.

Cons

  • Narrower than a general AI — not useful outside PM docs.
  • Template-driven output can feel uniform across teams.
Best for: PMs drafting and reviewing specs Pricing: Paid from $5/mo chatprd.ai

Linear AI

AI features baked into Linear — triage, draft issue descriptions, suggest owners, auto-link PRs.

Pros

  • Integrated — no tool-switching.
  • Reduces triage load meaningfully on busy teams.
  • Linear's pace of improvement is fast.

Cons

  • Only useful if you already live in Linear.
  • AI suggestions occasionally mis-assign complex threads.
Best for: teams already on Linear Pricing: Included with Linear paid plans linear.app

Notion AI

AI layer across your Notion workspace — summarise, draft, Q&A across pages, database fills.

Pros

  • Useful for synthesising long-lived workspace knowledge.
  • Database AI fills are quietly transformative for structured work.
  • Cheap add-on for existing Notion users.

Cons

  • Output quality lags behind dedicated tools like ChatPRD for specs.
  • Q&A across pages is only as good as your workspace organisation.
Best for: teams with a big Notion workspace Pricing: $10/user/mo add-on notion.so

Productboard AI

Ingests feedback from tickets, sales calls, and interviews; analyses trends; suggests what to build next.

Pros

  • Genuine feedback-to-roadmap consolidation.
  • Trend detection flags themes before humans would notice.
  • Strong integrations with common CX and support tools.

Cons

  • Enterprise-priced — out of reach for smaller teams.
  • Automation can over-flatten nuance; needs human review.
Best for: mid-market and enterprise product teams Pricing: Enterprise tiers; contact sales productboard.com

Granola

Meeting intelligence — joins calls silently, transcribes, produces structured notes and action items.

Pros

  • 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 meetings Pricing: Free tier + paid from $18/mo granola.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 studies Pricing: Paid from $30/user/mo dovetail.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 testing Pricing: Free tier + paid from $99/mo maze.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 testing Pricing: Paid from $75/mo syntheticusers.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 practices Pricing: Enterprise; contact sales usertesting.com