CodeIntegrity raises $5M to put permanent monitoring wires on unpredictable AI agents
Computer security startup CodeIntegrity on Wednesday announced a $5 million seed round to fund its efforts to build smart defenses for AI applications.
The San Francisco-based company made headlines last year when it showed how easy it was to manipulate AI models at many tech companies to share confidential information, earning a mention in The Economist last September for compromising a note-taking app in less than four hours.
“Every company was aiming to introduce agents to deployment, and they didn’t know how to do it securely,” said Abi Raghuram, CodeIntegrity’s founder and CTO.
The core challenge is fundamental. Conventional software relies on deterministic controls, meaning that when you type “X,” the computer always does “Y.” But AI agents – tools that can perform computer tasks automatically – are not deterministic because they are driven by natural language models. That makes them vulnerable to “quick injection” attacks, where a bad actor injects malicious text into a model and triggers the agent to do things like expose sensitive data.
To keep these agents in check, companies can hire people to supervise or issue a second LLM as a judge, but neither method is too dangerous or completely unreasonable.
“There are going to be a lot of agents that will be used in business situations, and no one has thought about (security) yet,” said Steven Jung, founder and CEO of CodeIntegrity. “We want to be the first to give these companies strict control.”
CodeIntegrity’s solution is to install a permanent guardrail called the runtime control layer. It acts as both a translator and a filter, forcing the unpredictable AI model to play by strict, predictable rules and restrictions on what business processes and data the AI agent is allowed to touch.
Raghuram and Jung met at the Rose-Hulman Institute of Technology, a top undergraduate engineering school in Indiana. After a few years of working separately at other employers, the two reconnected and founded CodeIntegrity in the Seattle area in May 2024. Later that year, they participated in a residency program at Antler in New York for their startup.
The now five-person company still has an engineer in Washington state, though the two founders recently moved to San Francisco.
Cybersecurity-focused Syn Ventures led the new round, with participation from existing seed investors Antler and Boost VC. The startup has raised $5.25 million in total.
About six companies are currently testing the CodeIntegrity product. The co-founders say they haven’t finalized their pricing model, but expect to offer annual contracts similar to those common in the cybersecurity industry.
Other startups are tackling agent AI security, including Seattle’s Certiv, which emerged from bankruptcy in March, and California’s Raven and Manifold Security.
Raghuram, who has worked at Seattle’s Truveta for more than three years, said he is eager to hire engineers locally, but called San Francisco “ground zero for the whole LLM thing.”
“Seattle is a strong place to hire great talent from Microsoft and Amazon, which is kind of what you want for a fast-moving startup,” he said. “But as founders, if you want to know what’s happening in the agency world, the Bay Area is the place.”
