Galileo AI (design tool)
Last reviewed
Jun 4, 2026
Sources
16 citations
Review status
Source-backed
Revision
v1 · 1,773 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 4, 2026
Sources
16 citations
Review status
Source-backed
Revision
v1 · 1,773 words
Add missing citations, update stale details, or suggest a clearer explanation.
Galileo AI (at the domain usegalileo.ai) was a generative AI design tool that turned plain-text prompts into editable, high-fidelity user interface designs, a workflow it popularized as "text to UI." It was built by a San Francisco startup of the same name, founded in 2022 within the South Park Commons community by Arnaud Benard and Helen Zhou, and it went viral in 2023 through demos that generated polished mobile and desktop screens from a one-line description. The company raised a 4.4 million US dollar seed round led by Khosla Ventures and opened a public beta in February 2024. In May 2025 Galileo AI was acquired by Google; the team and its technology moved into Google, and the product was relaunched inside Google Labs as Stitch, a Gemini-powered UI design tool. This article is about the design tool. A separate, unrelated company also called Galileo (the AI evaluation and observability vendor at galileo.ai) is covered in the disambiguation note below.
Galileo AI sat at the front of a product team's workflow, before any code or detailed Figma file existed. A user described an app or website in natural language, for example "a friendly dashboard for dog walkers," and Galileo returned several visual options for the interface, complete with layout, UI components, placeholder imagery, and realistic copy, usually in under a minute. The output was meant to be a starting point rather than a finished asset: designs could be edited further with text prompts and exported to Figma for refinement. The founders described the tool in shorthand as a kind of ChatGPT for interface design, aimed at designers, founders, and developers who wanted to move from idea to a concrete interface quickly.
The product also accepted visual input. Through an image-to-UI feature, users could upload a sketch, screenshot, or wireframe and have Galileo produce a corresponding high-fidelity design, and it could generate multi-screen user flows rather than single screens.
Galileo AI was founded in 2022 in San Francisco, incubated within South Park Commons, a selective community and early-stage fund for technical founders. The two co-founders brought complementary backgrounds. Arnaud Benard, the chief executive, had been an AI researcher at Google (his time at Google Research is generally placed around 2018 to 2021), where he worked on natural language models and on prediction for the Android keyboard, and he had also worked at the e-commerce company Faire. Helen Zhou joined as a founding designer and brought product design experience from Facebook (now Meta) and the self-driving company Cruise. Benard led the modeling and infrastructure work while Zhou focused on encoding design principles into the system.
A recurring theme in the founders' early account was that they were building in an area with little prior art. Benard noted that, unlike image or text generation, "there are no existing benchmarks, papers, or models for generating UIs," so the team built its own models, datasets, and supporting infrastructure largely from scratch. Rather than relying on an off-the-shelf model, they trained specialized natural language processing models on tasks aligned with design principles in order to produce usable, high-fidelity layouts.
Galileo AI launched a private beta in October 2023, sending out its first batch of invitations from a waitlist that had built up around its demos. The private beta drew strong interest from the design and founder communities. On February 6, 2024 the company opened the tool to a public beta, removing the waitlist gate, and adoption was immediate: the founders reported that the community created more than 50,000 designs within the first 24 hours, and that the tool reached roughly 100,000 design generations in under three days.
The company's public funding was a single disclosed round.
| Round | Date announced | Amount | Lead investor | Notes |
|---|---|---|---|---|
| Seed | February 6, 2024 | 4.4 million US dollars | Khosla Ventures | Announced alongside the public beta |
The seed round was led by Khosla Ventures and also drew a group of angel investors prominent in design and software, including Julie Zhuo (former vice president of design at Facebook/Meta and co-founder of Sundial), Airtable chief executive Howie Liu, David Hoang, Dan Becker, and Sana Labs founder Joel Hellermark, among others. Public databases such as Tracxn list this 4.4 million US dollar seed as the company's only disclosed funding event prior to the acquisition. (Some secondary write-ups cite a slightly higher rounded figure or additional pre-seed support from South Park Commons; the consistently sourced headline number is the 4.4 million US dollar Khosla-led seed.)
While operating independently, Galileo AI used a credit-based subscription model layered on top of a free tier. Image generations consumed credits while text exchanges were free. According to contemporary reviews and the company's pricing page, the paid tiers were a Standard plan at 19 US dollars per month (around 1,200 credits) and a Pro plan at 39 US dollars per month (around 3,000 credits, roughly 300 designs, plus a private mode). These prices applied to the standalone product and became moot after the move to Google, where the successor tool was offered free within Google Labs.
On May 20, 2025, at Google I/O, Google announced that it had acquired Galileo AI. Benard confirmed the news directly, posting that "Galileo AI has been acquired by Google" and that the team had "launched today the next generation of our product, powered by Gemini: Stitch." Zhou likewise framed it as a continuation of the work she and Benard had started in 2022. Coverage characterized the deal primarily as a talent-and-technology acquisition: the founding team joined Google to continue building generative design tools, and the standalone usegalileo.ai service was wound down, with existing users given a short transition window (reported as around 30 days) to migrate their work.
The successor product, Stitch, launched the same day as an experiment inside Google Labs. Google described it as a tool that turns "simple prompt and image inputs into complex UI designs and frontend code in minutes," powered by the multimodal capabilities of Gemini 2.5 Pro. Stitch carried forward Galileo's core ideas and extended them in directions the original tool had not covered:
| Capability | Galileo AI (standalone) | Stitch (Google Labs) |
|---|---|---|
| Text to UI | Yes | Yes |
| Image / sketch to UI | Yes | Yes |
| Front-end code export | No (design output only) | Yes (HTML/CSS and front-end code) |
| Figma handoff | One-click export to Figma | "Paste to Figma" |
| Underlying model | Custom in-house models | Gemini 2.5 (Pro and Flash tiers) |
| Pricing | Free tier plus paid plans | Free within Google Labs, with monthly generation limits |
A notable difference is that the original Galileo AI produced designs but did not export working code, a limitation Stitch addressed by generating front-end code directly. Google has continued to develop Stitch after the acquisition, adding real-time, conversational "vibe design" features in which a Stitch agent streams its work to a canvas and lets the user steer iterations as they happen.
Galileo AI was one of the earliest tools to demonstrate convincing text-to-UI generation, and its viral 2023 demos helped popularize the idea that a usable interface could be conjured from a sentence. It arrived alongside a broader wave of generative design and design-to-code tools and is frequently cited as a precursor in that space. Its short, roughly three-year arc from a South Park Commons project to a Google acquisition is often used as an example of how quickly the generative design category moved, and of Google's strategy of acqui-hiring small teams to seed its own AI products. After the acquisition, much of the public discussion of "Galileo AI" shifted to its successor, Stitch, which inherited the user-facing concept while running on Google's own large language models.
Galileo AI the design tool should not be confused with Galileo at the domain galileo.ai (formerly branded Rungalileo), a separate and unrelated company that builds an enterprise platform for evaluating, monitoring, and guardrailing large language model and AI agent applications. That company was founded in 2021 by Vikram Chatterji (chief executive), Atindriyo Sanyal, and Yash Sheth, raised a 45 million US dollar Series B led by Scale Venture Partners in October 2024 (about 68 million US dollars in total funding), and in April 2026 was the subject of an announced acquisition by Cisco for integration with Splunk's observability products. The two companies share only a name; their founders, funding, technology, and acquirers are entirely distinct. Everything else in this article refers to the design tool acquired by Google.