Audit is the part of corporate life almost no one thinks about, and almost everyone pays for. Public companies spend thousands of hours a year collecting evidence, testing controls and writing it all up. Most of it is done by hand, in Excel, much the way it was in 2002.
Andera wants to hand that work to AI. The San Francisco startup, founded in 2024, has raised a $37mn Series A led by Lightspeed Venture Partners to automate internal audit from end to end.
A reasoning problem, not a rules problem
Audit resisted automation for a reason. An auditor has to read messy, multi-format evidence, judge whether it satisfies a control, and document it well enough to survive a regulator. No rules engine could do that.
Large language models can try. Andera’s platform reads financial evidence across spreadsheets, PDFs, screenshots and journal entries, then renders an audit judgment for controls under Sarbanes-Oxley and other regimes. “The models are already sufficiently smart,” chief executive Aryo Patel said. “The hard problem is getting from billions of tokens to the 30,000 that matter.”
