Software engineers who have spent years perfecting code review workflows may find those skills becoming secondary sooner than expected. The newly installed CloudBees CEO Mo Plassnig sees a future where human oversight of every line of code simply collapses under its own weight. Plassnig took the helm at CloudBees earlier this month, replacing Anuj Kapur.
His background includes founding Codeship, a continuous integration and delivery platform CloudBees acquired in 2018, and serving as chief product officer at the data security firm Immuta. He now faces a tech industry where the sheer volume of code generation has broken old assumptions.
The ability to write code is no longer a constraint, Plassnig said. End users will soon express intent directly, and that intent gets converted into prompts that generate code and even a pull request. The bottleneck shifts from writing software to reviewing and deploying it.
Current DevOps workflows still depend heavily on software engineers to review code. That code is being created at an overwhelming pace.
The math does not work. Human review scales linearly. AI-generated code scales exponentially.
Something has to give. Plassnig does not argue for an overnight rewrite of engineering practices. He describes a gradual transition.
Existing tools, platforms, and processes will be re-engineered over time. Software engineering teams will move toward what he calls agentic engineering workflows.
AI agents will take over tasks humans currently perform: reviewing code, observing system behavior, and deploying changes. This shift touches every company building software. The way applications are built and deployed is undergoing a fundamental transformation, Plassnig noted.
DevOps workflows must evolve to accommodate code from both professional developers and citizen developers — people who are not trained programmers but can now generate functional code through AI tools. For CloudBees, the implications are direct.
The company’s platform sits at the intersection of continuous integration and continuous delivery. If those workflows change, the platform must change with them. Plassnig’s appointment signals that CloudBees intends to lead that re-engineering rather than follow it.
For software engineers, the message is uncomfortable but clear. The job of reviewing every pull request will not scale. Engineers will need to shift from gatekeeping code to defining the rules and guardrails under which AI agents operate.
The work becomes less about spotting syntax errors and more about specifying what acceptable behavior looks like. For citizen developers, the barrier drops further.
Express intent, get code. No need to understand branching strategies or deployment pipelines. The agent handles that.
This opens software creation to a much wider pool of people within organizations. The consequences for security and governance are not fully addressed in Plassnig’s outline.
When AI agents deploy code automatically, who owns the failure? When citizen developers generate pull requests without understanding the underlying infrastructure, who audits the result? These questions hang over the transition.
What to watch next: whether CloudBees ships specific agentic engineering features in its platform. Plassnig built Codeship from scratch. He knows the difference between a vision statement and a product roadmap.
The industry will watch which tools get re-engineered first and how fast engineering teams adopt them.
