Basis Raises $100M Series B at $1.15B Valuation, Redefining What AI Agents Can Do for Accounting
Basis just closed a $100 million Series B round at a $1.15 billion valuation, and this one lands with real weight behind it. The round was led by Accel with Miles Clements, alongside GV and Lloyd Blankfein, while existing backers Khosla Ventures doubled down through Keith Rabois and Vinod Khosla. This isn’t a speculative AI story built on demos and deckware; it’s a company already embedded in the operational core of accounting firms, quietly reshaping how real work gets done, hour after hour, engagement after engagement.
What makes Basis interesting isn’t just the size of the round or the valuation bump, but the timing. Accounting is under visible strain: chronic talent shortages, rising client expectations, shrinking margins, and a level of burnout that firms rarely talk about publicly but feel daily. Into that pressure cooker, Basis has inserted long-horizon AI agents that don’t just assist humans but actually execute entire workflows across CAS, Tax, and Audit. Roughly 30% of the Top 25 accounting firms are already using Basis agents, which is a staggering adoption signal for a category that’s traditionally conservative and risk-averse.
The investor roster reads like a cross-section of modern tech power. Nat Friedman and Daniel Gross via NFDG, Box CEO Aaron Levie, Quora CEO Adam D’Angelo, Replit’s Amjad Masad, former Stripe COO Claire Hughes Johnson, Hugging Face CEO Clem Delangue, Jeff Dean from Google, Jeff Wilke from Amazon Consumer, Scott Belsky from Adobe, and Noam Brown from OpenAI all participated. That mix matters. It signals belief not only in AI, but in agents that persist, reason, coordinate, and finish work over long stretches of time rather than firing off single-shot answers and calling it a day.
At the product level, Basis is borrowing a page from the AI coding world, but pushing it further into regulated, high-stakes territory. Their agents operate continuously in the background, coordinating tasks, pulling documents, reconciling data, and returning completed deliverables for human review. The recent demonstration of an AI agent autonomously completing an end-to-end 1065 tax return feels like a quiet line-crossing moment. Not flashy, not viral, but deeply consequential for an industry built on precision, trust, and review cycles that haven’t fundamentally changed in decades.
Miles Clements at Accel framed it bluntly: teams that get the fundamentals right tend to pull away quickly, and Basis is years ahead in accounting AI. That assessment lines up with how Basis works closely with foundation model providers, including OpenAI, to stretch what current models can reliably do in economically meaningful environments. Prashant Mital from OpenAI described Basis as a frontier partner, pushing models into real-world accounting tasks that demand accuracy over hours, not seconds. That distinction matters more than most AI hype cycles want to admit.
There’s also something quietly telling about how Basis runs itself. The company has built an internal group called Atlas, dedicated to deploying agents across its own engineering, sales, and talent functions. It’s an agent-native philosophy turned inward, a kind of dogfooding that suggests they genuinely believe this is how organizations will scale under time pressure in the years ahead. Keith Rabois summed it up with characteristic intensity, describing Basis as an early blueprint for the agent-native organization of 2030, executing with unusual clarity and speed because it has to.
Vinod Khosla’s prediction may sound bold, but it fits the pattern: 20–50% efficiency gains across accounting practices today, followed by a step-change similar to what software engineering experienced in 2025. If that holds, accounting won’t disappear, but it will change shape. More review, more judgment, more client-facing work, less grinding through workflows that machines can now handle without fatigue or resentment.
The new capital will go toward deepening the platform, expanding engineering and ML teams, and pushing agents into even more complex workflows. Matt Harpe, CEO and co-founder, keeps the framing grounded. The goal isn’t replacing accountants; it’s equipping them with AI that actually works, improves quality of life, unlocks growth, and lets firms say yes to work they currently have to turn away. It’s a practical ambition, maybe even a little unglamorous on the surface, which is often how the most durable technology stories begin.
