Google’s former security leads raise $13M to fight email threats before they reach you

As AI is helping growing hackers To launch mass-scale email attacks, Google security leaders have joined the force to create autonomous AI agents, which target phishing, malware and business emails with threats before reaching user inboxes.

That’s the mission behind AugiceA new email protection startup that has just derived from stealth to Essel and Foundation Capital’s co-leading seed funds with $ 13 million.

More 90% of successful cybertacks begin CISA per US Federal CyberCsis Agency, including phishing email. A crowdedstreak recently Study (PDF) also learned that phishing messages produced by large language models (LLM) were 54% click-thro rate in 2024, which is much higher than 12% rate for human-written email.

To fight this increasingly threatening threats with AGCE’s autonomous AI agents.

Established by former Google Safe Browsing and Recaptcha Executive C Sai Khormai and Ryan Luo, the startup provides an orchestrated network of real-time AI agents autonomously inspecting, analyzing and neuting e-mail threats without relying on any specific rules. This method challenges the general email protection platforms that depend on the fixed rules and often require extensive training of the user.

“The sum of all the evil is that an email attachment to an email is that it always started where all the attacks started and so I really wanted to solve this problem,” said Khormei in a private interview with TechCrunch.

Khoramai was the head of the Google Product Management Product and Director of Product Management for more than five years until July 2021, that at that time, he led Google, his four billion users and four billion users for the production of four million websites, and web, and web, and web. It was at this time that he first met Luo, who spent about a decade in Google and was part of the Safe Browsing team.

He told TechCrunch that Google had given first experience in creating the detection technology, the deepest perception of protection from the company’s point of view and how to develop and scale the fast security business, he told Techchen.

Prior to Google, Khurmi Sales Detective Platform established the contmed, which was Earned by SugarcRM in 2016The He later served as VP for managing the product for more than a year and a half until November 2021.

AGCA has created a rational agent, each of which is a custom-Bilt LLM with a specific threat. Once the orchestrating agent recognizes any threats or possible threats, it calls other agents on the network, referring to Khormai “friend”. These agents then run analysis, argue with each other and respond to the orchestrating agent with the judgment.

Agents perform real-time analysis of each message components with links, attachments, metadata, QR codes and behavioral patterns.

“What we know from making these tools in Google is all the things we need to analyze what are all the data sources? What are all data sources? What are all the strategies to spot the attack, and all the staff opponents have seen more than 10 years of chess game with opponents?” Kharmai said.

Although AGCA has currently created more than 10 agents for this purpose, Khormai told TechCrunch that over 50 to 100 agents may have over time because opponents become smart and try to fool the system.

“I fully believe that opponents within two years will understand what we are doing they will rebuild what we are doing and then we have to make more agents to stay ahead of them,” he said.

Unlike a general email protection platform, which uses a rules-based approach, these AI agents look for a number of attacks and self-sources for every possible variant of those attacks in real-time, Khoormai said. Startup has created several threats and several AI models created for specific industries, including Venture Capital and Financial Services.

In addition to detecting threats, agents of AGCAs help to reduce false positivity up to 90% of the Traditional Solution, demanding startup.

For customers, Google Workspace or Microsoft 365 Email account does not take “more than five minutes” for customers to install AGCI system. Once it is set up, the startup will send a report within a few days with details about what the system is available in the environment with false positive and false negative. It will then only run in the read-mode for a week and then activate the separation.

“It is so difficult to solve these different problems in the email without this technology in the email.”

Startups, including San Francisco and New York offices, are currently running a pilot with customers in the United States and Europe, and already added three paid subscribers with data privacy compliance software locker and crypto payment platform fake connection. There is currently a six -member team in the startup.

With the new investment, Khormai said that the startup was planned to expand its technical skills and create a powerful cow-to-market infrastructure.