Skip to content

CodeIntegrity

Runtime guardrail platform for production AI agents, built by offensive-security researchers who publicly compromised the Notion app in under four hours.

Visit Website ↗ + Add to Compare
55/100Incremental Innovator

Overview

CodeIntegrity builds runtime guardrails for AI agents deployed in production, aiming to give engineering and security teams a way to constrain what an agent can actually do at execution time rather than relying solely on prompt-level instructions or after-the-fact monitoring. The company positions its approach as protection that persists even when an agent is manipulated or behaves unpredictably, rather than protection that depends on the agent behaving as instructed.

Founded in San Francisco in 2024 by CEO Steven Jung and CTO Abi Raghuram, CodeIntegrity built its early reputation on offensive security research, including a widely cited exercise in which the founders compromised the Notion application in under four hours — work referenced by The Economist in 2025. The company announced a $5 million seed round in May 2026, led by SYN Ventures with continued participation from Antler and Boost VC.

CodeIntegrity’s offensive research background gives it more independently notable technical credibility than many AI-agent security peers at a similar funding stage, even though its guardrail platform itself has not yet undergone public independent testing.

Innovation Matrix Assessment

Innovation Velocity 6/10

Moved from offensive-research notoriety (the Notion compromise) to a funded, productized runtime guardrail platform within about two years of founding.

Operational Value 6/10

Runtime guardrails for production AI agents address a real operational gap for engineering teams shipping agentic features faster than they can secure them.

Market Momentum 5/10

A $5M seed from SYN Ventures, Antler, and Boost VC is a modest but credible early round for a two-year-old company.

Category Disruption 5/10

Runtime guardrails for agentic AI are a needed evolution of application security, but CodeIntegrity competes with a fast-growing field of similarly positioned AI-agent security startups.

Real-World Efficacy 4/10

The founders' Notion compromise is independently notable, credible offensive-research evidence of skill, though it demonstrates the team's capability rather than independently validated efficacy of the CodeIntegrity product itself.

Enduring Relevance 7/10

As more production applications embed AI agents, runtime-level protection against unpredictable agent behavior is likely to stay a persistent need.

Why CISOs Should Care

It gives application security teams a way to put hard limits on what a production AI agent can do, even if the agent itself is manipulated or misbehaves.

What Makes It Different

Its guardrails are enforced at runtime and hold even when an agent is compromised, rather than relying on the agent following its instructions correctly.

The Matrix Verdict

55/100 — INCREMENTAL INNOVATOR

CodeIntegrity is an Incremental Innovator: genuinely credible offensive-security pedigree (the Notion compromise, cited by The Economist) applied to a real and growing problem, still awaiting independent validation of the product itself.

Editorial Note: Claims vs. Verified Findings

Founders, funding, and the Notion compromise research are independently reported (GeekWire, The Economist reference, TheSaaSNews). Specific guardrail efficacy claims for the commercial product are vendor-sourced and not yet independently tested.

Sources