Startup Cymphony raises 30 million dollars to secure AI agent access to company data
Startup Cymphony has raised 30 million dollars in funding (including 25 million in Series A) from Sequoia Capital and SMBC Fin Atlas Beyond Fund for a platform that monitors AI agent access to company data. Its post-investment valuation exceeds 100 million dollars.
Cymphony, a startup developing a platform for monitoring AI agent access to company data, has raised 30 million dollars in funding. This includes a 25 million dollar Series A round co-led by Sequoia Capital and SMBC Fin Atlas Beyond Fund, and follows a previously undisclosed seed investment from Sequoia Capital. The post-investment valuation of the startup exceeds 100 million dollars. Cymphony is based in New York and Tel Aviv and has been operating for two years; it was founded by Shy Dekel (CEO), Idan Berkovits and Edi Gotlieb, all of whom went through the Israeli military program Talpiot.
The platform, called “workforce graph", is intended to give security teams a unified view of employees, AI agents and other non-human identities, including the systems and sensitive data they can access. It combines identity, data and activity signals. According to the company Dekel, the motivation is the fact that enterprise security was designed for human employees, while AI agents operate as independent entities without comparable access controls.
According to Cymphony, at one public U.S. company it identified approximately 85 000 files to which AI tools and agents had gained access; the company says it helped close the exposure and verified that the files had not been accessed through these AI systems. In another case, according to the company, an external contractor deployed an unapproved instance of the Claude model from Anthropic, which used his existing access to search thousands of sensitive files.
Cymphony says it has dozens of enterprise clients, including KKR, Syngenta, Cass Information Systems and Athennian, and achieved annual recurring revenue in the millions of dollars in its first year of sales. According to sources, the market for AI agent security is crowded — Microsoft, Okta, CyberArk, Wiz and Varonis also address similar risks.
Why it matters
AI agents routinely receive access to the same company systems and data as employees, but without comparable identity and permission controls — and unlike humans, they can change their capabilities, acquire new permissions or create additional agents while running. This creates blind spots that, according to the cited cases, can lead to the exposure of tens of thousands of files or unauthorized searches of sensitive data using someone else's access.
Relevant practical impact
What this means
For individuals
According to Cymphony, in one case an external contractor deployed an unapproved instance of the Claude model from Anthropic, which used his existing access to search thousands of sensitive files — deploying an AI tool yourself under company credentials can therefore cause it to go through far more data than the user intended.
Check the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.