Michigan’s Swept AI raises $1.4 million for AI safety

Saginaw, Michigan startup Swept AI has raised $1.4 million in a pre-seed funding round aimed at scaling its platform for supervising and stress-testing autonomous AI systems.

The round was led by Chicago-based venture firm M25 and included participation from Wellington Management Company, BuffGold Ventures, Ann Arbor SPARK, Service Provider Capital, The Unicorn Group and angel investors. Swept said in a statement that the capital will be used to hire additional staff and further develop the product.

Swept’s product is designed to audit, interrogate and continuously supervise agentic AI: systems that act autonomously, make decisions and learn over time. The company said that as these AI agents are deployed in higher-stakes settings such as finance, healthcare and cybersecurity, traditional testing and monitoring tools are inadequate for handling behavior and compliance concerns.

“We help businesses take control of unreliable AI before it ever reaches their customers,” co‑founder and president Amy Fox said in their statement, going on to say that the company’s used “math, not vibes to prove the trustworthiness of AI systems ” 

The market for agent supervision is emerging as enterprises and policy makers seek ways to manage risks posed by increasingly autonomous systems and policy. The European Union even passed the AI Act to create a set of “risk-based rules for AI developers and deployers regarding specific uses of AI”

Mike Asem, founding partner at M25, described Swept’s approach to the “nondeterministic reality of agentic AI” as likely to become critical infrastructure for high‑risk, high‑value use cases.

Swept positions its platform as distinct from conventional model-monitoring tools by emphasizing adversarial evaluation and continuous oversight of agents in production. The company markets its service to enterprise security teams and AI vendors that need to demonstrate safety and reliability. 

According to data gathered by consultancy firm EY earlier this year, 76% of surveyed companies were using or planning to use agentic AI within the following 12 months, although only 56% of them were familiar with the associated risks.

Based on the fact that consumers were on average more than twice as worried as the executives surveyed about AI-related concerns, startups offering validation, monitoring and compliance tooling could see more demand as enterprises prepare for both commercial deployment and regulatory scrutiny.

Swept’s raise adds to a growing number of early‑stage companies seeking to create tooling around the operationalization and oversight of AI. How rapidly customers adopt such platforms seems to depend on factors including integration with existing workflows, the effectiveness of automated testing and the evolving regulatory landscape.