Google Stitch's Free AI Design Tool Has Market

When Google bundles a capable product into its free tier, that's not just a feature announcement. It's a market structure decision. Keep that in mind as we work through what Google's latest Stitch update actually does — because the product story and the competitive story are running in parallel, and only one of them is getting told.

Stitch is Google's AI design tool, available free at stitch.withgoogle.com. You describe a website or app interface in plain English — or out loud — and it generates screens, layouts, and user flows. That much existed before. What changed at Google I/O, according to SEO content creator Julian Goldie's walkthrough published this week, is the workflow architecture. The old version worked in turns: prompt, wait, review, repeat. The new continuous streaming model abandons that entirely.

"The Stitch agent now renders UI components directly onto the canvas as you type or speak," Goldie explains, "reflowing layouts in real time before a generation is even complete. You're no longer waiting for it to finish before you can correct it. You steer it while it builds."

That's a different interaction paradigm, not just a faster one. The distinction matters because it changes what skill is actually required. Directing a design as it materializes in front of you is closer to conversation than to specification-writing. The barrier to entry drops again.

Stitch can now generate up to five interconnected screens simultaneously — landing page, feature page, testimonials, sign-up, confirmation — with consistent typography, colors, and components across all of them. Finished designs publish directly to the web via Netlify integration, without a developer in the loop. Real-time multiplayer editing, launched alongside the streaming model, lets multiple users work the same canvas simultaneously.

All of this is free. No paid tier. No per-seat pricing.

The Figma question no one is asking officially

Goldie notes that Figma's stock declined sharply following an earlier Stitch update this year, citing a 12% drop over two days. I want to be precise here: I cannot independently verify that specific figure or its causal attribution, and readers should treat it as a market signal worth investigating rather than a settled fact. What is documented is that Figma went public and that Google's Stitch ambitions have registered as a competitive threat in the analyst community.

The more interesting question — the one that sits squarely in my lane — is whether anyone in a regulatory office is examining the structure of what Google is doing. Google is offering a functional equivalent to Figma's core collaborative design product, powered by Google's own AI infrastructure, distributed through Google's existing product ecosystem, and priced at zero. Figma charges per seat per month. Stitch charges nothing.

This is the classic pattern that draws attention from competition regulators: an incumbent with dominant distribution using that distribution to undercut a specialized competitor on price. The EU's Digital Markets Act was specifically designed to address this kind of leveraging behavior among designated "gatekeepers." Whether Google holds gatekeeper status under DMA in a way that would cover Stitch is a legal question I'd want an EU DG COMP attorney to answer — but the question is worth asking, and I haven't seen it asked anywhere. The FTC, under whatever posture it currently holds toward Big Tech, has the same structural concern available to it.

I'm not asserting antitrust liability. I'm noting that the competitive dynamics here have a shape that regulators have cared about in other contexts, and the silence from that direction is notable.

design.md: An open-source format that Google controls

The most underreported aspect of this update is design.md, described in Goldie's walkthrough as a format "open-sourced in April 2026, which captures your full design system — typography, colors, spacing, and component rules — in a format that AI tools can read and apply consistently." The cross-tool compatibility claims — Claude Code, Cursor, GitHub Copilot — are sourced from Goldie's demonstration rather than independently verified documentation, and readers building workflows around this should confirm directly with Google's release notes.

But take the architecture seriously, because design-to-code bridging has been the stubborn unsolved problem in AI design tooling, and design.md is Google's proposed solution. The format solves a genuine pain point: AI-generated designs drift. Each generation looks slightly different. Your brand erodes incrementally until nothing matches. A standardized machine-readable design system file, applied consistently across every generation, addresses that directly.

Here's what concerns me about how this is being discussed. "Open-source" is doing a lot of work in Google's framing. Publishing a spec under an open license is not the same as developing it through an open governance process. The W3C exists because the web needed standards bodies that weren't controlled by single vendors. IETF exists for the same reason. If design.md becomes the de facto standard for communicating design systems to AI tools — and its integration with Cursor and GitHub Copilot suggests Google is actively pushing for that outcome — then a critical piece of AI-adjacent infrastructure will have been specified unilaterally by the company that also controls the primary tool implementing it.

That is a standards play, and it should be evaluated as one. The open-source label does not resolve the governance question. I'd argue design.md warrants involvement from something like the W3C's AI standards working groups, or at minimum a formal multi-stakeholder process before it solidifies into critical infrastructure. Google writing the spec and Google building the tool that implements the spec is not an arrangement the industry should accept quietly, regardless of how useful the spec turns out to be.

The sameness problem in AI design is real, and design.md addresses it. That doesn't mean Google should be the sole author of the solution.

The workforce dimension no one has a policy answer for

Goldie's framing is entirely and honestly from the business owner's perspective: "That used to take a design team." He means it as a selling point, and for his audience — small business owners trying to build landing pages without agency budgets — it is one. But the people who were the design team are worth naming.

The mid-market freelance design economy — the independent contractors, the small studios, the solo UX practitioners doing conversion optimization and landing page work for clients who can't afford Figma Enterprise or an in-house design team — that's the segment that absorbs this kind of displacement first. Not the senior product designers at FAANG companies, and not the offshore production shops already operating at price floors. The middle.

There is no meaningful policy response being prepared for this. The EU AI Act addresses high-risk AI systems; a design tool generating landing pages doesn't meet that threshold. The US has no federal AI workforce framework at all. The policy gap isn't new — it's been the persistent failure of AI governance discussions to reckon seriously with labor market effects in skilled-but-automatable work — but Stitch's particular combination of capability, price point, and distribution makes it a clean case study in how that gap compounds.

What this actually is

Goldie's walkthrough — enthusiastic, product-focused, aimed at business owners who want to move faster — is a useful demonstration of what Stitch can do operationally. The dependencies this creates deserve equal time alongside the capabilities it unlocks.

Google has built a genuinely capable AI design tool and made it free. The product works, at least at the demonstration level Goldie is showing. The streaming model solves a real workflow friction. The multi-screen generation is legitimately useful for anyone building conversion funnels. The publish-to-Netlify integration removes a real bottleneck.

And Google has done all of this in a way that positions it to own a design infrastructure standard, undercut a direct competitor on price without any near-term revenue pressure to moderate that behavior, and absorb a portion of the professional design market into its ecosystem without meaningful regulatory scrutiny.

Those things are all true simultaneously. The FTC and EU DG COMP should be curious. The W3C or an equivalent body should be in the design.md conversation. And anyone advising freelance designers on career positioning right now should be honest about what "that used to take a design team" means when the person saying it works for Google.

Samira Okonkwo-Barnes is Buzzrag's tech policy and regulation correspondent.