Arden, a newly launched Y Combinator startup from Spring 2026, is automating internal audit and SOX compliance testing using AI agents that pull evidence from company systems, run control tests, and produce audit-ready documentation in minutes instead of weeks. The platform targets a massive but underserved market: every public company in the United States is required to conduct SOX testing, yet most teams still rely on spreadsheets and manual screenshots to prove their financial controls are trustworthy.

Why Is SOX Compliance Such a Burden for Public Companies?

SOX compliance, formally known as the Sarbanes-Oxley Act, requires public companies to test hundreds of internal controls annually to demonstrate that their financial reporting is reliable. The process is labor-intensive and expensive. According to Arden's founders, SOX compliance costs nearly $2 million per year at a typical public company, with over 70 percent of that expense coming from labor costs. When internal audit teams cannot keep up with the workload, companies outsource overflow work to consulting firms at rates between $500 and $1,000 per hour, adding another seven figures to the annual bill.

The core problem is that audit work has remained largely unchanged for three decades, even as the number of systems companies must audit has doubled. Auditors spend weeks chasing colleagues for evidence buried across 20 or more disconnected systems, manually testing controls in Excel, and repeating the same work every quarter. Existing audit software stores completed workpapers but does not automate the testing itself, leaving the most time-consuming part of the job untouched.

How Does Arden's AI-Powered Approach Work?

Arden is built as an AI-native platform that runs audit tests end-to-end, connecting to a company's control library, source systems, and historical audit data. The platform learns how each audit team works and builds that knowledge over time. The system operates through several key capabilities:

• Evidence Gathering: Computer-use AI agents navigate legacy systems and capture screenshots that external auditors require, while direct integrations pull data from modern software stacks, eliminating weeks of back-and-forth communication.
• Automated Follow-Up: A Slack or Teams agent follows up with control owners to confirm evidence, saving auditors weeks of chasing down approvals while providing control owners with pre-filled context.
• Control Testing: Arden's testing engine supports both full-population testing and sample-based testing, catching every exception and flagging issues for human review.
• Audit-Ready Documentation: The platform produces workpapers with full traceability, tying every conclusion back to source evidence in a format external auditors can immediately use.

The founders, Aryaman Khanna and David Lomelin, bring deep expertise in AI systems and machine learning. Khanna previously worked on AI infrastructure at Databricks and studied computer science and data science at UC Berkeley, where he co-authored research on privacy-preserving machine learning for autonomous systems presented at NeurIPS 2025. Lomelin previously worked on AI at MIT and co-founded Blueprint, conducting research at MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) on large language model post-training and applied AI systems.

What Makes This a Significant Opportunity for Y Combinator?

Arden's entry into Y Combinator's Spring 2026 batch reflects a broader trend of AI startups targeting enterprise workflows that have resisted automation for decades. The internal audit market is massive but fragmented, with thousands of public companies and pre-IPO firms facing identical compliance challenges. By automating the most labor-intensive parts of SOX testing, Arden addresses a pain point that generates immediate, measurable cost savings for customers.

The startup is actively seeking introductions to VPs of Internal Audit, Chief Audit Executives (CAEs), and SOX directors at public companies, as well as CFOs and controllers at public or pre-IPO firms. The team is also targeting Big 4 accounting firms and mid-tier consulting firms that perform co-sourced SOX work, suggesting a go-to-market strategy that spans both direct enterprise sales and partnerships with established audit service providers.

Arden's approach represents a shift in how AI startups are thinking about enterprise automation. Rather than building general-purpose tools, the company is designing AI agents specifically for the workflows, systems, and regulatory requirements of a particular function. This specialization allows the platform to deliver immediate value without requiring customers to fundamentally change how they work.