By BasisLearning Objectives
- Understand what "agentic accounting" means and how it differs from an accounting chatbot
- See how an AI agent can produce finished, reviewable work such as a tax return
- Evaluate why a human accountant's review and professional responsibility remain essential
What Is Basis?
Basis is an agentic artificial-intelligence platform built for professional accounting firms. Rather than answering questions like a chatbot, Basis deploys AI agents that actually perform accounting work — reviewing documents, reconciling accounts, and preparing returns — and hand back finished output for a human accountant to check. The concrete problem it solves is capacity. Accounting firms face a well-documented talent shortage and mountains of repetitive, labor-intensive work; Basis takes on the multistep grind so scarce human experts can focus on review, judgment, and client relationships. In one notable demonstration, a Basis agent completed an end-to-end partnership tax return — the United States Form 1065 — from source documents through to a finished workbook ready for professional review.
The company, Basis, is headquartered in New York and was founded by Matthew Harpe and Mitchell Troyanovsky. In a 2026 Series B round led by Accel alongside GV (Google Ventures), it raised roughly 100 million dollars at a valuation of about 1.15 billion dollars. By that point its agents were in use at a meaningful share of the largest United States accounting firms, positioning Basis as a flagship anchor for the emerging category of agentic accounting.
💡Key Concept
Agentic AI that does the work: A chatbot tells you how to do a task; an agent does the task. Basis strings together the many steps of an accounting job — pulling data, applying rules, preparing schedules — and delivers a completed deliverable rather than advice. The accountant's job shifts from producing the work to reviewing and signing off on it, which is where their professional judgment adds the most value.
✅Tip
Visit Basis: getbasis.ai — built for professional accounting, tax, and audit firms; enterprise deployment, pricing by quote.
Core Capabilities
Autonomous multistep work
Basis agents execute the sequence of steps a real accounting task requires — gathering documents, reconciling figures, preparing schedules — rather than handling a single isolated prompt, so the output is a finished piece of work.
Tax return preparation
The platform can take a return from source documents through to a completed, reviewable workbook, demonstrated on an end-to-end partnership return (Form 1065). The finished return is prepared for an accountant to review, not filed autonomously.
Works alongside accountants
Basis is positioned as a teammate rather than a replacement: it produces draft deliverables that human accountants review, correct, and approve, keeping the professional firmly in control of the final product.
Firm-scale deployment
Built for professional firms rather than individuals, Basis is designed to slot into how accounting practices operate, which is why it has been adopted across a number of the largest United States firms.
Strengths
- Produces finished work, not just answers: Agents deliver completed deliverables for review, addressing capacity directly rather than offering guidance.
- Targets the profession's real pain: It attacks the repetitive, high-volume work squeezing understaffed firms.
- Demonstrated on complex tasks: Completing an end-to-end partnership return shows ambition beyond simple data entry.
- Backed and adopted at scale: A 2026 Series B led by Accel and GV, plus use across major firms, signals serious traction.
Limitations & Considerations
- A human accountant reviews and signs. Basis produces work for review; the accountant must check it and remains professionally responsible for what the firm delivers and files. The AI does not carry that accountability.
- Accuracy and hallucination risk. Like any AI system, agents can make errors or produce plausible-but-wrong output, so rigorous human review is essential — especially on returns and filings where mistakes carry real consequences.
- The firm keeps professional responsibility. Regulatory and ethical duties to clients and tax authorities stay with the licensed professionals; delegating the labor does not delegate the responsibility.
- Enterprise, firm-scale tool. Basis targets professional firms and involves deployment and workflow integration, not a quick self-serve sign-up for an individual.
Best Use Cases
| Task | Why Basis |
|---|---|
| Preparing returns for review | Produces a finished workbook, demonstrated on a partnership Form 1065 |
| Clearing repetitive accounting work | Agents handle multistep reconciliation and document review |
| Expanding firm capacity | Takes on labor-intensive tasks so staff focus on judgment |
| Reviewing rather than producing | Delivers draft deliverables for accountant sign-off |
Getting Started
- Identify the high-volume, repetitive workflows in your firm — reconciliations, document review, return preparation — where agents can help most.
- Engage Basis to scope a deployment that fits your firm's systems and professional review processes.
- Establish a firm review standard so every AI-produced deliverable is checked and signed by a responsible accountant.
- Pilot on a defined engagement type, verifying accuracy closely, before broadening the agents' scope.
Key Takeaways
- Basis is an agentic AI platform for professional accounting firms that produces finished work — including tax returns — for a human accountant to review.
- It demonstrated an AI agent completing an end-to-end partnership return (Form 1065) and is used across a number of the largest United States firms.
- Founded in New York, it raised about 100 million dollars in a 2026 Series B led by Accel and GV at roughly a 1.15-billion-dollar valuation.
- The honest caveats are that a human accountant reviews and signs the work and the firm keeps full professional responsibility.
