Oxus Wants to Make Internal Auditors 10x Faster With AI

The Macro: Internal Audit Is a $20 Billion Industry Running on Manual Labor

Internal audit is one of those functions that every public company and most pre-IPO companies must have, and nobody talks about. SOX compliance alone requires companies to document, test, and verify hundreds of internal controls every year. The work involves reading thousands of pages of documentation, creating process flowcharts, testing controls against evidence, and generating workpapers that auditors and regulators can review.

Most of this work is done by humans with spreadsheets. The Big Four accounting firms charge premium rates for it. Smaller audit shops use a combination of AuditBoard, Workiva, and ServiceNow GRC. But these tools are mostly workflow management platforms. They organize the work. They do not do the work.

The market is massive and growing. SOX compliance costs publicly traded companies millions per year. Pre-IPO companies increasingly need audit infrastructure before they can go public. And the labor shortage in accounting means qualified auditors are expensive and hard to find. Anything that can compress audit timelines without sacrificing quality has an enormous buyer pool.

The Micro: Three MIT Grads Taking on the Big Four

Janet Liu, Melinda Liu, and Christine Watts founded Oxus. All three studied at MIT. Janet has a background in digital payments research at the Federal Reserve and investment strategy at D.E. Shaw. Melinda built AI products for fintech and worked in finance at JPMorgan. Christine built ML systems for startups and did consulting at BCG and NERA. The combination of AI expertise and financial services experience is exactly right for this product.

Oxus automates four core audit workflows. Control Testing generates test procedures and workpapers automatically. Walkthrough Flowcharts convert process documentation into visual diagrams. SOX Scoping helps audit teams assess risk and scope their work. Document Intelligence lets auditors search, analyze, and annotate documents with AI assistance.

The differentiators they emphasize are evidence-grounded outputs tied to specific documents, an auditor-in-the-loop review process, and standardized formatting. These sound like small details, but in audit, they are everything. Auditors need to show their work. Every conclusion must trace back to a specific document. Outputs that look inconsistent or cannot be sourced get rejected by external auditors and regulators.

They are a three-person team from YC Winter 2026 working with Diana Hu. Current customers include payroll providers and public biotech companies.

The Verdict

Oxus is attacking a genuinely underserved market with the right team. Internal audit has been resistant to automation because the work requires judgment, not just processing. But the judgment layer is increasingly something AI can handle, especially when the underlying data is structured documentation with clear rules.

The competitive risks are real. AuditBoard is well-established and has the budget to build AI features. Workiva is publicly traded and also investing in AI. But enterprise software transitions are slow, and the incumbents have massive codebases that make AI integration harder than building from scratch. Oxus has a greenfield advantage.

In 30 days, I want to see completion rates. What percentage of control tests generated by Oxus pass external auditor review without modification? In 60 days, the question is whether the 10x speed claim holds up across different company sizes and industries. In 90 days, I want to know the pipeline of pre-IPO companies. If Oxus can position itself as the standard audit tool for companies preparing to go public, the growth trajectory is extremely strong.