This Startup Secretly Detects Fraud For Fortune 500s
Y Combinator Startup Podcast Podcast Recap
Published:
Duration: 31 min
Summary
Variance, a company that emerged from stealth with a $21 million Series A, builds AI agents for risk and compliance to detect fraud and automate identity reviews at scale. Co-founders Karine and Michael leverage their experience from Apple to address fraud detection challenges for Fortune 500...
What Happened
Variance has emerged from stealth mode, announcing a $21 million Series A funding round. The company specializes in building AI agents that help Fortune 500 companies handle risk and compliance by automating content, fraud, and identity reviews at scale. This technology has been crucial for platforms like GoFundMe, where it is used to verify fundraisers and prevent fraudulent activities, especially during crises when fraud attempts spike.
The founders, Karine and Michael, met while working on fraud engineering at Apple. Their experience in the industry led them to create Variance, focusing on using AI to replace human analysts with intelligent agents. These agents use behavioral signals and context to verify fundraisers and handle complex identity reviews and KYB verifications.
Variance's AI agents are designed to connect to a wide range of internal and external data sources, including over 100 business registries and the open web. This connectivity is essential for tracing complex fraud patterns and ensuring comprehensive fraud detection capabilities. The company also handles the integration of unstructured data from multiple sources, which has been one of its biggest technical challenges.
The technology behind Variance evolves rapidly with LMS improvements, allowing for the development of self-healing systems that can adapt and respond to new fraud patterns. These AI agents have proven effective in detecting complex fraud rings, even identifying state-sponsored actors during elections.
Variance's team is relatively small, with 12 members, including 5 software engineers, and is based in San Francisco. The company is currently hiring more staff to meet growing demands. The team culture emphasizes strong ownership and in-person collaboration to drive innovation and effectiveness.
Karine faced a significant personal challenge when she was hit by a truck, yet continued to lead the company with resilience. During her recovery, Michael brought her a book on Norman Foster, symbolizing resilience and reflecting on the challenges and future of their company.
Variance's first customer was IAC, which used their technology to review marketing content for compliance. The founders have maintained a strong focus on solving specific problems in fraud detection, leveraging early large language models to automate tasks even before the advent of systems like ChatGPT.
The company's dedication to solving specific fraud detection challenges resonates deeply with its customers, who value the specialized expertise and innovative solutions that Variance provides. The founders' commitment to using their skills for the good of the industry highlights their sense of duty and purpose.
Key Insights
- Variance has successfully secured a $21 million Series A funding round to further its development of AI agents for risk and compliance in Fortune 500 companies. The company remains discreet due to the sensitive nature of its work with fraud and identity reviews.
- The company's AI agents have been instrumental in verifying fundraisers on GoFundMe, especially during crises when fraudulent claims increase. By using behavioral signals and contextual analysis, Variance replaces human analysts with more efficient and scalable technology.
- One of Variance's technical challenges is integrating unstructured data from diverse sources, which is crucial for tracing complex fraud patterns. Their AI agents can scrape data from human-oriented UIs and connect to over 100 business registries and the open web.
- The founders, Karine and Michael, bring their expertise from Apple to the startup, having a clear vision from the beginning on solving fraud detection issues. Their personal dedication to the mission is exemplified by Karine's resilience after a major accident and their strategic use of early LLMs before systems like ChatGPT.