Clark Raises $11M Seed for AI Audit Employees at CPA Firms

Client evidence has its own timetable, and an audit team inherits the work of remembering what is still missing. Clark is building AI employees around that interruption: ask for the document, keep track of the response and continue the assignment when the evidence turns up. The New York company announced an $11M Seed round on November 2, 2026, alongside the launch of its Clerks.

Touring Capital led the financing, with returning investors Marathon Management Partners, Rerail and Sofeon Ventures participating. Founder and CEO Enes Witwit is taking the business further into audit execution, where a professional has to review both the result and the evidence supporting it. Licensed CPAs retain judgment, review and sign-off on the audit opinion.

For accounting firms, the useful distinction is how much coordination comes back with the delegated work. A system that can follow an unfinished assignment through the arrival of new information could change an engagement team's daily workload. Clark's financing supports that proposition, along with hiring and broader engagement coverage, while the company works toward continuous monitoring between audits.

What Clark's Seed Round Finances

Clark says the new round brings total funding to $16M. There is a historical reporting difference worth keeping visible: the latest announcement describes a September 2025 round as $5M, while the original uiAgent release announced $4.6M. The $16M figure is Clark's current reported total; the available releases do not explain that difference.

The capital will support senior hires in product, engineering and sales, wider deployment across firms and engagement types, and efforts to help CPA firms extend services into government auditing. Clark's careers page describes an on-site New York team spanning technical, audit and commercial roles. That mix matters because an audit product has to enter a firm's working habits, including how reviewers challenge evidence and how clients supply it.

Clerks are assigned to engagements to review documents, perform tests and prepare workpapers. Clark says they also request missing information, track the response and resume work when it arrives. For a buyer, the value of that sequence lives in the handoff: how much context a team must reconstruct before the assignment can move again. Faster document processing alone would leave that coordination question unanswered.

Where an AI Employee's Assignment Begins and Ends

Clark describes Material Intelligence as the operating technology behind its agents. The current product page sets out a practical delegation model: define the client, reporting period, source records, working environment and firm methodology. The AI then completes the defined task within those boundaries rather than requiring an instruction for every step.

That product page still uses the name Ava for its AI teammate, while the funding announcement introduces Clerks. Both descriptions put software inside an engagement with defined work to perform. The important purchasing detail is the scope a firm can establish around that work, rather than an assumption that a new product name confers broader professional authority.

A reviewer needs to understand which period a test covers, which records it used and how the result supports the workpaper. Those are useful ways to evaluate the promised continuity. If an incoming document changes the relevant evidence, the team needs a clear account of what the software did with it and which matters still require judgment.

The audit opinion remains with licensed professionals. That responsibility makes reviewability part of the product's everyday usefulness: a team has to be able to examine the work while it is still possible to investigate an exception. Software that advances an assignment also needs to leave the people supervising it able to follow its progress.

A Guarantee That an Audit Firm Can Read Closely

Clark's peer-review guarantee gives the supplier relationship unusually concrete boundaries. The company says a covered extraction error or missed flag that becomes a peer-review finding can trigger damages compensation and a refund of fees paid on that client. Damages are capped under the agreement, and a firm's decision to override an item the AI flagged is excluded.

Those terms deserve attention beside the release's broader accuracy language. A guarantee establishes a commercial commitment for defined circumstances; it does not establish an independently measured zero-error rate. An accounting firm can examine what counts as a covered error, which engagement records support a claim and where its own professional decisions remain outside the promise.

That is a substantive buying conversation. An impressive workpaper demonstration still leaves questions about the file a reviewer will receive, the exceptions the system will surface and the record available if something goes wrong. Clark is offering a supplier commitment close to those questions, giving a firm more than a speed claim to assess.

Touring Capital co-founder and general partner Priya Saiprasad frames assurance as a trust business in the funding announcement. The investor's accompanying essay makes its own thesis explicit: AI can take on substantive execution while licensed professionals retain oversight and final sign-off. That is the investor's conviction about the opportunity, with the eventual operating results still belonging to each deployment.

From an Engagement Tool to a Continuing Relationship

Enes Witwit previously worked on machine learning at SAP before building Clark for the accounting profession. The company, formerly uiAgent, now reports that more than 35 CPA firms use its technology on audits. That customer count is company-reported and offers a measure of adoption, without supplying an independently audited picture of revenue or productivity across the customer base.

The next stretch of the business follows naturally from the work Clark wants to own. Adding firms means encountering different methodologies, evidence patterns and expectations about review. Expanding into more engagement types adds further demands on how an assignment is defined and how its output can be examined. Product development, customer implementation and the sales promise will meet in the same engagement file.

Continuous monitoring remains a development ambition in the announcement. Clark wants findings to identify what they cover and when they were last checked, extending useful verification between periodic audits. That would give an accounting firm another relationship to manage with its clients: the continuing currency of the evidence behind a finding.

For now, the $11M Seed finances the people and product work around that progression. As Clerks enter more assignments, the firm's reviewer will still need to follow the evidence from the client's latest document through the work performed and into the professional decision recorded in the file.