Making DevSecOps Work in the Enterprise: What’s Really Changing
by Patrick Morris
AI Is Rewriting the Rules of Security
Teams are experimenting with AI-driven, dynamic security controls that adapt in real time.

Think access that’s automatically granted when someone’s working on a relevant feature, and revoked the moment they’re done. No tickets. No manual admin. Just-in-time security that tightens controls without slowing people down.
Some organisations are going further with AI-powered code remediation. They’re moving beyond AI that simply flags issues to AI that can propose and sometimes implement secure fixes aligned with internal standards.
Retrieval-augmented generation (RAG) is proving key here, letting AI draw on a company’s own patterns rather than the noise of the open internet. While the group agreed it’s not fully mature, early trials are showing promising results.
One of the most immediate benefits is false positive triage. Several leaders shared how AI tools are cutting through noisy results, eliminating up to 40% of false positives. That’s more than just efficiency. It restores
trust in the tooling. Developers are far more likely to engage with a system that doesn’t waste their time.
As AI starts to generate and secure more code, the developer role is shifting. Developers are moving towards reviewing, prompting, and maintaining AI outputs. Skillsets are evolving from pure coding to prompt engineering, validation, and operational oversight.
The big question now: do teams have the right skills for this next phase, and how do they get there?
Ease of Use Drives Adoption
Everyone agreed that developer adoption makes or breaks DevSecOps.
Security tooling only works when it fits naturally into the development workflow. Developers will skip tools that slow them down, throw up cryptic errors, or demand context-switching.
But when security is embedded directly into the IDE or CI/CD pipeline, with clear guidance and useful feedback, adoption soars.
Some organisations are even accepting small compromises in accuracy to keep usability high. A tool that catches the critical 80% and fits seamlessly into the workflow is far more valuable than one that flags everything but no one uses.
Training also looks very different now. Long, static sessions are out. Teams are finding real traction with short, interactive formats, CTFs, code labs, or micro-demos, that show the real-world impact of vulnerabilities. Developers respond when they can see how a flaw could actually be exploited, not just when they’re told it’s “bad practice.”
Crucially, there’s a constant feedback loop between security and development. The best teams treat security as a living system, refining approaches based on what developers say works and what doesn’t.
Culture: The Hardest and Most Important Part
Culture came up again and again. The organisations that make DevSecOps work are those that build a shared mindset between development, operations, and security.
Security champions are a key part of that. These embedded practitioners act as translators, advocates, and problem-solvers within teams. Some operate as evangelists, others as hands-on contributors. The common thread is collaboration, cross-team champion groups are raising security maturity collectively, not just in isolated pockets.
A few leaders spoke about “amnesty conversations”, regular moments where teams share mistakes, incidents, and near misses without fear of blame. These sessions build trust and turn risk into learning.
Maturity roadmaps are also essential. The most successful programmes treat DevSecOps as multi-year transformations with clear milestones, not one-off initiatives. This structure gives teams the predictability they need to plan and measure progress.
Threat modelling remains a sticking point. Developers don’t always “think like attackers,” which can leave gaps. Embedding security specialists directly into engineering teams helps bridge that divide.
Some teams are also experimenting with AI-generated threat models or security-approved design patterns. These are useful accelerators, but everyone agreed that human oversight remains critical to ensure recommendations are practical and relevant.
Winning Executive Support
At the executive level, security is getting serious attention. But leaders still face a communication challenge: how do you prove the value of something whose success is defined by nothing happening?
Framing security in terms of risk appetite, resilience, and business continuity resonates far more than technical metrics alone. Boards respond when you talk about risk reduction and near misses, not just vulnerabilities and patch counts.
“Security’s success is invisible. Our job is to make the invisible visible.”
This quote from one participant captured the challenge perfectly.
Tooling: A Reality Check
Every team uses a different mix of tools. But the consensus was clear: no silver bullet exists.
Enterprises need layered, integrated solutions that balance coverage with usability.
Static analysis: Tools like CodeQL were praised for strong performance across multiple stacks, as long as they integrate smoothly into developer workflows.
Secret scanning: Remains essential but time-consuming. Without automation and discipline, it quickly turns into tech debt.
DAST: Brings valuable runtime insights, but only works if scanning cycles stay fast. Developers won’t tolerate tools that slow down commits or pipelines.
Some powerful tools were called out for poor developer experience and high false positives. Even the best features don’t matter if they interrupt delivery.
The takeaway: the most effective teams choose tools that are fast, embedded, and meaningful not just feature-rich.
The Bottom Line
If there’s one theme that ran through the discussion, it’s that developers do care about security, when it’s done right.
That means:
Fast, contextual tools: Keep security invisible until it matters.
Workflow-aligned security: Fit into developer processes, not around them.
Open, shared culture: Build trust and collaboration across teams.
Executive support: Link cyber risk directly to business impact.
With thanks to all the leaders who joined and shared their experiences, your insights made this conversation real and practical. We look forward to continuing it.
Interested in joining the conversation?
If you’re a senior security or technology leader and would like to be part of future peer-to-peer council discussions, reach out to amber.williams@riversafe.co.uk to express your interest in joining the Cyber or Technology Leaders Council.