# De · Design — element 23 of 58 > Where the product gets its face. Turns ideas into interfaces and identity. - **Group:** 5 · Creative & Design - **Necessity:** Essential - **Price band:** $ · under $30/mo - **Maturity:** Stable - **Edition:** v2026.Q3 · verified 2026-09-13 ## Leading tools (v2026.Q3) - **Figma** — the product-design os - **Canva** — non-designer speed at scale - **Framer** — the site design agent - **Penpot** — open-source, self-hosted - **Paper** — agent-native canvas ## Our take AI collapsed production design; it hasn't touched taste. This element is now judgment plus a very fast pencil. ## Combines with Im, Vb - Guide: [Best design tool for AI startups (2026)](https://elems.ai/best/best-design-tool-for-ai-startups.html) ## The top 5 — deep dossier (verified 2026-09-13) Figma, and for product UI it isn't close — post-IPO it's the industry's operating system, with Q2 2026 revenue of $370.1M (a third straight quarter of accelerating growth, ~39% guided for the year), Make prompt-to-prototype on every paid seat, and Config 2026 shipping Motion, Weave, and agent-facing MCP — even as the stock trades ~75% off its IPO-pop high. Canva is the pick when the founder is the designer: $4B ARR, 265M monthly users, and an AI suite used 800M times a month. Framer wins when the marketing site is the product's face; Penpot when self-hosting or zero lock-in is non-negotiable; Paper when agents, not humans, do most of your production design. 1. **Figma** (Figma, Inc. (NYSE: FIG)) — Free (500 AI credits/mo) · Pro full seat $16/mo · Org $55 · Enterprise $90; Dev $12–35, Collab $3–5; AI credits metered per tier. Best for: Any team designing actual software — UI, design systems, prototyping, and the design-to-code handoff, now with agents wired in over MCP. Why: The default professional canvas got an AI platform bolted on and kept growing: Q2 2026 revenue $370.1M with full-year guidance raised to ~$1.465B (+39%), 136% net dollar retention, 85% gross margin. Make (prompt-to-prototype-and-code) ships on every paid seat; Config 2026 (Jun 24) added Motion, Weave generative media, and code layers; Code Connect landed in the MCP server Aug 5, 2026, feeding agents real component context. Watch: The market is unconvinced: stock under $23 and ~$12B market cap by Aug 6, 2026 — down ~80% from the $56B IPO-pop peak — on AI-disruption fears (Anthropic board member Mike Krieger resigned Apr 2026 over competing design tooling). AI credits make heavy Make use a real line item, and the seat matrix is genuinely confusing. [https://www.figma.com](https://www.figma.com) 2. **Canva** (Canva Pty Ltd) — Free · Pro ~$15/mo · Teams ~$10/user/mo (min 3) · Enterprise custom — vendor page blocks verification; re-check. Best for: Founders and non-designers producing brand, marketing, social, decks, and now apps at volume — one subscription, zero design training. Why: The scale story of the category: ~$4B ARR by end-2025 (+43%), 265M monthly active users, 31M+ paid subscribers, and a $42B valuation set by an Aug 2025 employee stock sale. Its AI suite (Magic Studio, Canva Code 2.0, Visual Suite 2) was used 800M times a month by end-2025, up 700% YoY — and buying Serif let it make the pro-grade Affinity suite free on Oct 30, 2025. Watch: Identity: it's a content factory, not a product-design tool — UI/design-system work still leaves for Figma. Recognized revenue ($3.5B 2025) trails the $4B ARR headline, the long-teased IPO keeps slipping, and pro designers resent the Affinity freemium-with-Canva-AI shift. [https://www.canva.com](https://www.canva.com) 3. **Framer** (Framer B.V.) — Free · Basic $10/mo · Pro $30/mo per site + AI credits · Enterprise custom; extra editors $20/mo. Best for: Design-led marketing sites and landing pages — the public face of the product — designed, published, and now agent-maintained in one tool. Why: Owns the 'site as design surface' slice: $100M Series D at a $2B valuation (2025), ARR doubled to $50M in 2025 with a $100M target for 2026, and enterprise logos (Miro, Perplexity, Scale AI) now the majority of new accounts. Framer 3.0 made it agent-first — native Design, CMS, and Code agents plus external hooks for Claude Code and Cursor. Watch: Sits on the border with vibe-builders and website builders — Webflow still leads that market (~$200M ARR), and app-from-prompt tools attack from the other side. Credit-based AI pricing has already been reshuffled once ('AI credits, simpler plans, lower prices'). [https://www.framer.com](https://www.framer.com) 4. **Penpot** (Kaleidos (open source, MPL)) — Free up to 8 members · Unlimited $7/user/mo (capped $175/mo) · Enterprise $25/user/mo · Private Server $50k/yr · self-host free. Best for: Lock-in-averse teams and regulated orgs that want a real Figma alternative they can self-host, with designs stored as open SVG/CSS. Why: The only serious open-source contender: 600,000 teams claimed, design files that are web standards under the hood (SVG/CSS, not a proprietary format), and 2026 moves upmarket — a $25/user Enterprise tier with SSO/audit logs, a $50k/yr private-server offering, and 'AI workflows with any agent/LLM' on every tier rather than a locked-in model. Watch: No native prompt-to-design generation — it integrates agents rather than shipping its own; polish, plugin depth, and prototyping still trail Figma. Vendor-reported user counts are the only adoption numbers available. [https://penpot.app](https://penpot.app) 5. **Paper** (Paper Design, Inc.) — Free (unlimited editors, 100 MCP calls/wk) · Pro $16/editor/mo annual ($20 monthly, 1M MCP calls/wk) · Org custom. Best for: Teams whose production design is done with agents — an HTML/CSS-native canvas that syncs tokens, styles, and components straight into the codebase over MCP. Why: The purest bet on agent-era design: the canvas is HTML/CSS, so what you draw is what ships, and agents (Claude Code, Cursor, Copilot) read and write it over MCP — no translation layer, no handoff. A $34M Series A from Accel and ICONIQ (Jun 2026) and a desktop app mark it as the breakout new entrant of the cycle. Watch: Young: small surface area next to Figma (vector editing still on the roadmap), no track record at enterprise scale, and no published adoption numbers. Betting on it is betting the agent-native workflow wins before Figma's MCP work closes the gap. [https://paper.design](https://paper.design) ### How to choose - If You're designing actual software — screens, flows, a design system, engineers downstream → Figma Professional ($16/seat). Make covers prompt-to-prototype, and the MCP/Code Connect path feeds your coding agents real components; budget for AI credits at heavy use. - If The founder is the designer and the output is brand, marketing, social, and decks → Canva Pro (~$15/mo) — 800M monthly AI uses says the Magic Studio workflow works; add the now-free Affinity suite when you need pro-grade vector/photo work. - If The marketing site is the product's face and you iterate on it weekly → Framer Pro ($30/site) — design, publish, and let its agents (or your Claude Code) maintain it; skip the Figma-to-webflow handoff entirely. - If Compliance, data residency, or lock-in aversion rules your stack → Penpot — self-host free or pay $25/user for the enterprise build; your files are SVG/CSS you actually own. - If Agents do most of your production design and code is the source of truth → Paper (MCP-native canvas) — or Onlook/Subframe if you want the canvas living directly in your React repo. ### The field (19 more) Sketch, Play (dead), Paper, Subframe, Magic Patterns, Onlook, Visily, Uizard (acquired), Stitch, Galileo AI (acquired), Relume, UXPin, Lunacy, Affinity (by Canva), Adobe Express, Adobe XD (fading), InVision (dead), Motiff (dead), Weavy (acquired) Full dossier data: https://elems.ai/e/de.json --- Source: [elems.ai](https://elems.ai/e/de.html) — the periodic table of the AI-led startup. Data: https://elems.ai/elements.json (CC BY 4.0, cite elems.ai).