Audit is becoming an AI-native workflow

Andera has raised a $37 million Series A led by Lightspeed, with participation from Bain Capital Ventures, A*, and Pear VC, to build AI agents for audit work. Lightspeed describes Andera as the company building an AI-native platform to automate internal audit, a function that still relies heavily on manual evidence collection, control testing, Excel workpapers and documentation.

The funding matters because audit is not an obvious automation category. It is high-stakes, evidence-heavy and deeply tied to trust. A weak output is not just inefficient; it can create regulatory, financial and reputational risk.

That is why the market signal is important: AI is moving from horizontal copilots into enterprise workflows where reasoning, documentation and defensibility matter.

The hard problem underneath audit

Lightspeed’s investment thesis frames audit as a reasoning problem, not a rules problem. An auditor has to read complex evidence, judge whether it satisfies a control objective and produce documentation that can withstand regulatory scrutiny. According to Lightspeed, this is why audit has resisted automation for decades.

Andera’s own framing goes in the same direction. The company says it is bringing together auditors and engineers to solve the harder problem underneath audit: reasoning over messy evidence the way an experienced auditor would.

That messy evidence can include Excel workbooks, PDFs, screenshots, journal entries and financial documents. Lightspeed says Andera processes hundreds of millions of tokens of financial evidence per control and applies AI to identify the right documentation and render an audit judgment.

Why Lightspeed is betting on Andera

Lightspeed says it spent months building a thesis around audit before investing. The firm spoke with finance and audit leaders across Fortune 500 companies, evaluated early-stage companies in the space and consulted Big Four partners, former PCAOB officials and practitioners. Its conclusion: audit is at a pivotal moment.

The pressure comes from two directions.

First, audit teams are under cost and efficiency pressure. Lightspeed says audit teams at public companies spend thousands of hours collecting evidence, testing controls and producing workpapers, much of it manually and often in Excel.

Second, AI has finally reached a point where it can take on parts of the workflow that older rules-based systems could not. Lightspeed points to large language model reasoning, long context windows and agentic tool use as the technical shift that changed the calculation.

What Andera is building

Andera builds agents that automate manual audit work, turning weeks of testing into minutes, according to the official post provided by the company.

Lightspeed says Andera automates control testing across SOX, operational and compliance controls, including user access provisioning, bad debt reserve controls and management review controls. The firm also says Andera’s architecture is built to reflect how experienced auditors work, including reading, processing and writing back into Excel dynamically.

This detail matters. Audit automation is not only about summarizing documents. The product has to move through the actual workflow: gather evidence, test controls, evaluate exceptions, create workpapers and leave behind a trail that audit teams can trust.

The market signal

Andera’s Series A points to a broader shift in enterprise AI.

The first wave of enterprise AI focused on assistants. The next wave is vertical agents that can handle specific high-friction workflows inside finance, legal, healthcare, compliance, security and operations.

Audit is an especially strong test case because the work is repetitive but not simple. It requires judgment, traceability and domain expertise. If AI can help auditors reason across messy evidence and produce defensible documentation, it could reset expectations for other back-office functions.

The bigger question is not whether AI can replace every auditor. Lightspeed’s thesis is closer to extension than replacement: AI handles evidence collection, testing and workpaper generation so audit professionals can spend more time on risk, judgment and enterprise integrity.

What to watch next

The next test for Andera will be trust at scale.

Audit teams will want accuracy, explainability, traceability, integration with existing systems and confidence that outputs can hold up under review. Large enterprises will also care about security, workflow fit and whether AI-generated workpapers can match the standards expected by internal and external stakeholders.

Andera’s opportunity is clear: audit has remained too manual for too long. But the bar is high because this is not a lightweight productivity tool. It is AI entering one of the enterprise’s trust functions.

If Andera can prove that agents can reason over evidence like experienced auditors, audit may become one of the strongest examples of vertical AI moving from promise to production.