Paper raises US$34M from Accel and ICONIQ for agentic design
Paper, a San Francisco design platform built for what it calls the agentic era, announced a US$34M Series A with Accel and ICONIQ. Participating investors include Designer Fund, Michael Grinich of WorkOS, Anton Osika of Lovable, and engineers and designers from Anthropic and OpenAI. Founded by Stephen Haney, who previously co-founded Modulz, the company behind Radix UI, Paper raised a US$4.2M seed from Accel in 2025 before this round. The capital will fund product development, team expansion, and deeper investment in AI-native workflows for designers. For investors and operators, the hook is the traction number underneath: this is a Series A priced on measurable commercial momentum, not on a thesis alone.
That momentum is the strongest data point in the announcement. Paper has grown ARR 25 times since the early 2026 launch of Paper Desktop and has appeared on Ramp's Fastest Growing Companies list for four consecutive months. Its customer list reads as a who's who of companies building modern software, including Ramp, Lovable, Vercel, PostHog, Quartr, and Y Combinator. ICONIQ partner Mariano Payano pointed to the adoption pattern specifically, noting that quality-obsessed teams are pulling the product in from all directions and that this kind of organic word of mouth is rare. When a category incumbent is as entrenched as Figma, the way a challenger grows matters as much as how fast, and that distinction drives the analysis below.
What Paper actually does
Paper is a design environment built so that designers and engineers work in a shared language, with AI agents participating directly in the creative process rather than sitting outside it. The core product development came with Paper Desktop, launched in early 2026, which turns the design canvas into a shared workspace for humans and AI agents, connecting design, code, and data in real time. The company positions itself against a specific gap: AI has transformed how software gets built, but design workflows have largely stayed the same, leaving designers working with tools built for a different era while being asked to move at agentic speed.
The mechanism behind the differentiation
The technical decision that defines Paper is that it renders with HTML and CSS instead of a proprietary canvas. Every element placed on the canvas is actual web code, which means designs connect directly to production code and to the agents increasingly responsible for writing it. The practical consequence is the elimination of the handoff gap: where a traditional tool requires translating a proprietary component format into production code, Paper's canvas is already a DOM environment, so copying a component yields usable code with nothing to convert. Paper has also built an MCP server with bidirectional tools, which lets agents not only read a design but modify it, populate frames with real data, and generate variations from natural language. As Haney framed the ambition, the goal is helping designers do deep, meaningful work at a pace that supports their businesses, working seamlessly alongside coding agents like Claude Code and Codex.
Market context
The round lands in the middle of the 2025 to 2026 reordering of software tooling around AI agents. The incumbent context is significant: Figma is a public company that raised roughly US$749M across seven rounds before listing, and its canvas is a proprietary environment where moving work to production requires a translation step. Paper's bet is that the translation layer itself becomes a liability once agents are doing the building, a structural argument rather than a feature comparison. Accel partner Dan Levine framed it as category creation, arguing that most design tools were built for a world where humans did all the work while Paper is built for one where humans and AI create together. The endorsements come from people building the surrounding stack: Vercel CEO Guillermo Rauch said Paper arrived at the right moment to channel intelligence on demand, and Tailwind CSS creator Adam Wathan credited the team with starting from first principles rather than inherited assumptions. The caveat for readers is that this is a company-issued announcement with no disclosed valuation, absolute revenue figure, or customer count, and a 25x ARR increase from a small base is a different signal than the same multiple from scale.
What this means for investors and business owners
- A platform shift reopens closed categories. Design tools looked settled until agents changed who consumes the output. When the underlying technology changes what a product is for, incumbency stops being protection and becomes inherited assumptions. The transferable question for any operator is which of your category's fixed rules exist because of a constraint that AI has quietly removed.
- Architecture is the moat, not features. Paper's advantage is not a better canvas; it is choosing HTML and CSS so there is no translation step at all. Features get copied in a quarter, while foundational architecture decisions take years to unwind. When evaluating a challenger against an entrenched competitor, look for a structural choice the incumbent cannot easily replicate without rebuilding, because that is where durable advantage lives.
- Adoption quality is a leading indicator. Ramp, Vercel, Lovable, PostHog, and Y Combinator adopting a young product signals something a raw user count cannot: technically demanding teams selecting it voluntarily. ICONIQ's partner specifically flagged the organic pull as rare. For founders and investors, tracking who adopts, and whether they came inbound, often predicts durability better than how many.
- Read growth multiples against their base. A 25x ARR increase is genuinely strong, but it is reported without a starting figure, so the absolute revenue remains unknown. Multiples from small bases are common and still meaningful, yet they are not comparable to the same multiple at scale. The discipline is to always ask what the denominator is before pricing a growth rate, whether you are assessing an investment or your own internal metrics.
- Investor composition is strategic information. Beyond Accel and ICONIQ, this round includes the founders of WorkOS and Lovable plus engineers from Anthropic and OpenAI. That is a cap table of people who build with these tools daily and who bring distribution into the exact community Paper needs. When raising, the composition of the round can be worth more than the terms, because aligned operators convert into early customers and credibility.