Every few years, cybersecurity reaches a point where familiar methods begin to strain under new realities. The shift to the cloud changed how teams thought about infrastructure. DevOps blurred the boundaries between development and operations. Now, artificial intelligence is challenging both — redefining how software is written and deployed, and exposing new kinds of security risks in the process.
Snyk’s launch of Evo enters that conversation as an attempt to address what many see as the next frontier of application security: defending AI-native and “agentic” software that can operate, reason and act on its own.
From Static Scanners to Adaptive Defenders
For years, application security tools have largely been reactive — built around static scans and post-deployment patching. That model assumes humans write code in predictable cycles. But AI-assisted and AI-generated development doesn’t follow that rhythm. Code can now be created, tested and deployed continuously, leaving traditional scanning tools struggling to keep pace.
Snyk describes Evo as a system that applies the OODA loop — Observe, Orient, Decide, Act — to embed security within the AI development process itself. Rather than functioning as a single scanner, Evo is designed to coordinate a series of specialized agents for discovery, threat modeling, red teaming and remediation through what the company calls a “workflow agent.”
Katie Norton, research manager for DevSecOps and software supply chain security at IDC, said this reflects a broader market trend. “The emergence of agentic security solutions like Snyk Evo signal a meaningful shift in the application security landscape,” she explained. “Traditional tools have focused on scanning, policy enforcement and compliance across predictable software systems; agentic solutions instead aim to make security itself autonomous and adaptive.”
While that vision remains aspirational for most organizations, it underscores a growing desire for security tools that can operate continuously and contextually, rather than through discrete checks and dashboards.
Policy Written in Plain English
One of Evo’s central claims is its use of natural language for policy creation. Security teams can describe access controls or usage restrictions in plain terms — such as “block unverified models from accessing production data” — and have those translated into executable rules.
If the feature works as intended, it could help make governance more accessible to teams without deep security expertise, though the extent to which this approach can scale across complex environments remains to be seen.
Securing the AI-Native Development Frontier
The evolution toward agentic systems is creating new challenges for developers and defenders alike. Melinda Marks, cybersecurity practice director at Omdia, said her research shows this concern is rising sharply. “AI is the top element of concern for application security and cloud security teams, and this is a growing challenge as AI technology and its usage rapidly evolves,” she said. “Snyk Evo leverages AI to apply it to the defenders’ side to secure development using AI, helping application security teams support AI-enabled development and AI-native applications.”
Marks added that Evo’s orchestration model — designed to work across multiple vendors — could help teams set and enforce policies earlier in the software development lifecycle, potentially reducing risks before deployment.
The Roadblocks Ahead
Analysts agree that embedding security within AI systems will not be simple. Many enterprises still lack visibility into how AI is used across their environments, from datasets to deployed models.
“Enterprises will struggle to operationalize agentic or embedded security until they achieve basic AI readiness,” Norton noted. “Most are still trying to gain visibility into where and how AI is used. Success will depend not only on internal maturity but also on strong collaboration with vendors.”
That complexity extends to governance. Agentic systems integrate deeply with data pipelines and runtime environments, requiring new forms of coordination and shared responsibility between technology providers and customers.
The Next Layer of Standardization
As agentic tools become more common, interoperability may determine their long-term success. Norton said the market is likely to see new frameworks emerge — similar to efforts like the Open Cybersecurity Schema Framework — to help these systems work together. “Some shared frameworks will almost certainly be needed as agentic systems evolve,” she said, citing early attempts such as the Agent-to-Agent Protocol, Model Context Protocol and IBM’s Agent Communication Protocol.
Each proposes ways for autonomous systems to exchange context and enforce policies across different environments, but none has yet gained broad industry consensus.
Where the Market Is Headed
Whether Evo delivers on its goals will depend on how effectively Snyk can turn its architecture into measurable outcomes for users.
Den Jones, founder and CEO of 909Cyber, said he welcomes the innovation but remains cautious. “At 909Cyber we’re always watching how the security landscape changes with the rise of AI — and the announcement of Evo by Snyk is an exciting step forward,” he said. “The more we can automate and orchestrate security in a smart, secure way, the better we can help our clients confront the new risks brought by AI-native applications and agentic systems. That said, the real value will lie in delivering actual substance, not just marketing momentum — we’ll be tracking closely to see that the capabilities live up to the promise.”
Agentic security is still an emerging idea, and Evo’s launch highlights how early the field remains. But it also reflects a growing consensus that static approaches to software defense won’t be sufficient for systems that think and build on their own. The next chapter of application security may hinge not on faster scanners, but on systems intelligent enough to understand the code they protect.