Basis raises $100M to automate accounting workflows as 300K accountants exit the profession
Feb 25, 2026 with Matthew Harpe
Key Points
- Basis raises $100M to deploy AI agents that automate accounting workflows like tax returns and reconciliation, targeting a profession losing 300,000 workers with 75% of remaining accountants facing retirement.
- The startup just released an AI capable of completing entire corporate tax return workbooks end-to-end, shifting accountants from doers to reviewers of agent output.
- Unlike law or coding, accounting faces structural talent scarcity that makes automation about survival rather than margin compression, since low prestige means few new entrants will compete on price.
Summary
Basis, a New York-based startup, raised $100M to scale its AI agent platform for accounting teams. The company automates accounting workflows—tax return completion, reconciliation, audit preparation—by deploying long-horizon agents that sit on top of existing ERP and accounting software rather than replacing it.
Matthew Harpe, Basis co-founder, positions the company's agents as a response to a structural labor crisis in accounting. Roughly 300,000 accountants left the profession during and after COVID, and an estimated 75% of remaining accountants will retire over the next decade with few new entrants replacing them. This shortage, Harpe argues, created urgency for automation that extends beyond typical AI enthusiasm—accounting firms need agents to handle work they otherwise cannot staff.
The review-not-do model. Basis's core pitch mirrors agent coding: accountants become reviewers rather than doers. The company just released an AI that can complete an entire corporate tax return workbook end-to-end, a capability Harpe says only emerged in recent months. Accounting naturally supports this handoff because the discipline already embeds multiple review layers to catch human errors. As accountants build trust in Basis's output on lower-risk tasks, they can reduce review intensity on routine work.
Harpe acknowledges accounting differs materially from coding or law. Accounting workflows are not text-in-text-out, and the internet contains sparse data on accounting procedures, making it harder for general-purpose LLMs to perform well out of the box. That gap has only recently closed as models improved in accuracy and complexity over the past year.
The context-handoff problem. Harpe sees persistent human judgment as unavoidable. Categorizing ambiguous expenses—is a charge CapEx or OpEx?—often requires context that agents cannot autonomously retrieve. Basis is investing in long-horizon agents that can request clarification over time rather than interrupt with a chatbot experience, balancing agent autonomy against the need to engage humans on genuinely ambiguous decisions.
On fee dynamics, Harpe notes that unlike law, accounting has not attracted fresh talent despite AI disruption. The profession's low prestige relative to law means it is unlikely to see an influx of new entrants competing on price, as legal has. Instead, accounting firms face a structural constraint: do more with fewer people or deploy automation. That constraint makes Basis adoption less about margin compression and more about capability preservation.