Cybersecurity startup CodeIntegrity announced a $5 million seed round on May 27 to build what it calls “permanent guardrails” for enterprise AI agents. The core problem the company targets: traditional software is deterministic, producing the same output for the same input every time. AI agents driven by language models are not, making them vulnerable to prompt injection attacks that can trigger data exfiltration or unauthorized actions.
“There’s going to be more and more agents being deployed in enterprise settings, and no one has figured [security] out yet,” CodeIntegrity co-founder and CEO Steven Jung told GeekWire. “We want to be the first one to actually provide that deterministic control for these companies.”
How the Runtime Control Layer Works
CodeIntegrity’s approach inserts a runtime control layer between the AI agent and the enterprise systems it interacts with. The layer acts as both a translator and a filter, enforcing strict rules about which systems and data an agent can access. Rather than relying on the model itself to behave correctly, the control layer provides deterministic enforcement regardless of what the model attempts.
This architectural choice reflects a growing consensus in enterprise security: you cannot trust the model to police itself. Prompt injection attacks, where malicious inputs manipulate an agent into performing unintended actions, remain an unsolved problem at the model layer. CodeIntegrity’s bet is that the enforcement must happen outside the model, at the infrastructure level.
The Competitive Landscape
CodeIntegrity enters an increasingly crowded agent security market. Seattle-based Certiv emerged from stealth in March 2026. California-based Raven and Manifold Security are also competing for enterprise agent security budgets. This week alone, Xage Security launched its Zero Trust for AI platform and Zscaler announced Project AI-Guardian with six global systems integrator partners.
The founders, Jung and CTO Abi Raghuram, met at Rose-Hulman Institute of Technology in Indiana. They position CodeIntegrity as a response to enterprise demand that emerged as companies began deploying agents built on frameworks like Claude Code, OpenClaw, and LangChain into production environments. “Every company was aiming to launch agents into deployment, and they didn’t know how to do it safely,” according to the company’s announcement.
Why Determinism Matters Now
The timing aligns with a broader market signal. Earlier this week, the Agent Control Standard announced a vendor-agnostic open standard for governing AI agents at runtime. BeSafe-Bench research published last week found that none of 13 production AI agents tested could complete 40% of tasks while respecting all safety constraints.
The $5M seed round will fund CodeIntegrity’s product development and early enterprise deployments. The company’s pitch to CISOs is straightforward: before you deploy agents into production, you need an enforcement layer that doesn’t depend on the model’s cooperation.