Figma buys Bud team to pull AI agents closer to the design canvas

Why it matters

Figma is using small AI team acquisitions to defend the design canvas as software creation shifts from static mockups to agent-built prototypes and apps.

Figma has acquired the team behind Bud, the Y Combinator-backed AI agent builder formerly known as Orchids, in a deal that gives Dylan Field's public design software maker another small technical team focused on turning prompts into working software.

TechCrunch reported the acquisition on July 7th, citing a post from Bud CEO Kevin Lu (@kevinlu625). Lu wrote on X that Figma is "one of, if not the, defining product companies of our time" and called it a "natural home" for the next phase of work. The purchase price and deal terms were not disclosed.

Bud's service life is ending quickly. Under the deal, both Bud and Orchids will shut down by July 18th, and users have been told to migrate projects before then, according to TechCrunch. That detail makes the deal easier to read: Figma is taking the people and product learning, while the standalone Bud and Orchids products are being wound down.

Lu and co-founder Bach Tran built Bud out of the 2025 AI application rush. Y Combinator's profile lists Bud as a Winter 2025 company based in San Francisco, founded in 2025, with a team of six. YC describes Bud as an "agent with a computer" that can work from web, text or Telegram, browse the web, build and code, create research, slides, documents and spreadsheets, and run autonomously. YC lists Lu as a former University of Pennsylvania computer science and math student with generative AI experience at AWS and research work at Stanford HAI; Tran is listed as a former Penn computer science student with engineering experience at Addepar.

Bud's arc also captures how quickly the AI app layer has been rearranged. Orchids began as a vibe-coding platform that let users spin up apps for mobile, web, Slack and browsers, according to TechCrunch. Bud later moved toward broader agent creation, with access to services, a browser and coding tools. Earlier this year, TechCrunch, citing a BBC report, said apps created on Orchids were susceptible to cyberattacks, a reminder that tools making it easy to generate software also inherit the security responsibilities of software distribution.

Figma is buying into the build layer

Figma has spent the last year pushing past its original identity as a collaborative design tool. The Bud acquisition sits in that pattern. Figma already sells Figma Make, which Figma describes as a way to prototype, polish and ship ideas, with code-backed work that stays visually editable. Figma's own product page says Make can use design systems, npm packages, Figma frames, PDFs and other attachments as context, then let users move between code and canvas.

The commercial reason is visible in Figma's filings and earnings materials. On May 14th, 2026, Figma said Q1 revenue rose 46% year over year to $333.4 million. Figma also said roughly 60% of paid customers with more than $100,000 in annual recurring revenue used Figma Make weekly during the three months ended March 31st, up from over 50% in the prior quarter. Those are Figma-reported figures, but they explain why the company is willing to keep acquiring around AI prototyping: Make is no longer just a demo feature inside a design product. Figma is using it as a wedge into software creation.

Figma has paired that product push with integrations into coding environments. In Q1, Figma said it expanded Code to Canvas support across tools including Claude Code, Codex, Cursor, VS Code and Warp, allowing interfaces generated in those tools to move into Figma as editable layers. Figma also said its Model Context Protocol capabilities let agents read and write directly to Figma files, creating or modifying design assets using components, variables and tokens.

That makes Bud's team relevant even if Bud's own apps disappear. Figma's hard problem is no longer generating a prototype from a prompt. Plenty of companies can do that. Figma needs the generated thing to remain legible to product teams, designers and engineers after the first prompt. Bud's work on agents that browse, use services and write code fits that direction, especially if Figma wants its canvas to become the place where AI-generated applications are edited, reviewed and handed off.

A public-company version of the AI talent grab

Figma is operating with a different set of incentives than the private AI coding startups chasing viral growth. Figma priced its IPO on July 30th, 2025, and its shares began trading on the New York Stock Exchange under the ticker FIG on July 31st. That public-market status changes how AI product bets are judged: Figma has to show that AI drives retention, seat expansion or usage in accounts that already pay meaningful software budgets.

Field has been explicit about the strategic framing. In Figma's Q1 earnings release, he said, "When code is a commodity, design is the competitive edge." It is a useful line because it explains why Figma is leaning into code while still defending design as the control point. Figma does not need to become a full IDE to benefit from AI coding. Figma needs to make the design file the shared surface where generated code, prototypes, motion, assets and product decisions converge.

Figma's June Config announcements pushed the same thesis. In a June 24th blog post, Field announced code layers, Figma Motion, shaders, generative plugins, Weave tools and a broader Figma agent. Figma said code layers would let users turn a design layer into an interactive code layer, duplicate it for alternate directions and collaborate around it on the same canvas. Figma also said its agent could connect to tools including Notion, Slack, Granola, Hex, GitHub and Atlassian.

The market is moving toward the same boundary from the other side. Coding products are adding design context, payments, deployment and workflow automation. In May, RuntimeWire reported that Visa backed Replit to explore agentic payments inside the IDE, a sign that developer tools are trying to become places where agents do work beyond code generation. Figma is approaching that fight from the product team's center of gravity: design files, components, brand systems and collaboration.

The Bud acquisition gives Figma a YC-backed team that has already lived through one fast pivot inside the prompt-to-software market. Lu and Tran started with generated apps, then moved toward agents with computers. Figma is making a related bet at public-company scale: the winning product surface for AI software creation may be the one where humans can still inspect, edit and govern what the agent made.