In tax and accounting, the hard part isn’t writing reports, it’s getting the right answer fast. The rulebook keeps growing, federal code, state and local ordinances, court opinions, treaties, accounting standards, and edge cases multiply with every new business model. The profession is stretched. Senior partners are retiring, juniors face steeper ramps, and the expertise gap widens. Teams spiral into escalations and burn hours on manual research, waiting for the one person who has seen it before. Clients feel the delay, and so do margins.
To solve that, Accordance launched today with a simple promise: give every professional an expert co-pilot that is steeped in real authority and tuned for the way firms actually work. The company is announcing $13 million in total funding led by Khosla Ventures and General Catalyst, with participation from Anthropic, NEA, Bain Capital Ventures, and Sequoia Capital.
The pitch is not a general chatbot. It is a domain-trained system that reads like a seasoned associate and cites like a careful reviewer. Founders David Yue and Finsam Samson, former fintech founders and Stanford AI researchers, say Accordance is already used by thousands of professionals, with 10x usage growth over the last six months as teams embed it in day-to-day work.
Complicated CPA workflows getting worse by the year
The practical problem inside firms is not just knowledge. It’s flow. On a Tuesday afternoon, a staffer may be staring at a multi-state nexus fact pattern, a deferred tax footnote under IFRS, and a client email asking about an election deadline, all while the senior reviewer juggles three audits and a partnership workout. Work slows for two reasons: facts are messy, and authority is scattered across systems. Researchers dig through subscription databases, PDFs, and local guidance, then attempt to synthesize and reconcile. Managers want a clear frame: what the rule says, how the facts line up, where the risk sits, and what to recommend. Without a tool that retrieves the right sources, reasons over them, and drafts a tight answer with citations, teams default to manual loops. That creates variable quality, uneven first-pass accuracy, and late nights before filings.
Accordance’s solution: the world’s most advanced AI for CPA teams
Accordance’s answer is a multi-agent system built for tax, audit, and accounting. The company says the platform is trained and evaluated on one of the most exhaustive libraries of authoritative federal, state, local, and international materials: statutes, regulations, revenue rulings, court decisions, treaties, standards, and evolving sub-national guidance. In plain terms, the system tries to do four things at once: pull the right authority for the specific facts, reason through competing interpretations, flag risks and edge cases, and produce a working draft, memo, client note, or workpaper comments, with linked citations so reviewers can verify. It is designed to decompose problems, check consistency against the corpus, and avoid shallow summaries that fall apart in partner review. The company claims it outperforms major foundation models on specialized tax and accounting tasks and says it ships updates as the rulebook changes.
Accuracy and trust are non-negotiable. The obvious risks, hallucinations, stale sources, and over-confident answers, are deal-breakers in this category. Accordance says it mitigates them with retrieval over vetted authority, reasoning steps that prefer conservative interpretations, and outputs that always show provenance. Firms also need governance. The product supports role-based controls, standardized templates, and review sign-offs that mirror existing QA processes. That matters for defensibility later, whether in audit files, tax controversy, or client audits of advisory work. The system also needs to live where teams already are: document repositories, ticketing, research databases, tax prep and audit suites. The company says integrations and export paths are a core part of adoption, not a nice-to-have.
Backed by top investors, founded by proven AI builders
Accordance also raised from some of the industry’s most notable investors who see a category shift rather than another tool.
“We’re witnessing the birth of a new category. AI that doesn’t just assist professionals, but fundamentally transforms how expertise is created and distributed.” said Vinod Khosla, Founder of Khosla Ventures. “Accordance doesn’t replace expertise, it multiplies it, enabling teams to transcend traditional limitations and reshape the entire landscape of accounting and tax services.”
This was echoed by Max Rimpel, a partner at General Catalyst, who added: “The Accordance team has deep technical skill and a rare grasp of the domain. They’re building trusted AI tools for some of the most vital and complex knowledge workers in the economy, and doing it with the precision and performance those users demand.”
It helps that the founders have strong AI research and founding experience. Yue and Samson studied at Stanford’s AI lab, built fintech products before this, and have set up research partnerships with OpenAI, Anthropic, and Stanford to track frontier improvements while keeping a tight domain focus. The company is also moving into education, with early rollout at the USF School of Law, to train the next cohort on workflows they will see in practice. For firms, that’s an adoption on-ramp: grads who already know how to use a domain AI with citations and review.
For now, Accordance is positioning itself as the top-of-the-stack expert rather than a background utility. The goal is not to replace judgment. It is to make expert-level analysis repeatable, reviewable, and fast. If the company can keep answers tied to authority, keep reviewers in control, and prove hard ROI in pilot cohorts, it will earn space in the core stack. If it can’t, firms will test it and move on. That’s the right filter.
Accordance is available now for tax, audit, and accounting teams at accordance.com.
