Variance

Variance builds AI agents that carry out risk investigations for regulated institutions, with anti money laundering as its main use. The agents take on casework from first line triage through to senior level reviews: they collect evidence, reason over it and reach a recommendation, such as escalating a case with the supporting artifacts. An investigation can draw on a company's internal transaction data and on external sources, including company registries, sanctions lists, adverse media, open web search and other open source intelligence databases. The agents extract information from unstructured material such as scanned or handwritten documents, images, websites and filings in other languages, and the product includes tools to verify identity documents and incorporation filings and to detect tampering. A context engine links entities across systems to show ownership structures and hidden relationships. Teams write their own policies and standard operating procedures in plain English, and the agents are meant to follow them case by case. Auditability is a central claim. Every decision carries an evidence trail, cited sources, step by step reasoning and configurable guardrails, aimed at compliance teams, regulators and auditors. The vendor states that its agents collect 90 percent of the evidence for each case. It also announced a $21M Series A. Pricing is not published. The site directs prospective customers to request a demo.