CloudBees lays out how AI is helping and hurting software development in the latest 2024 DORA DevOps report. The company confirms AI boosts productivity but cuts delivery stability by 7.2%.
The problem: AI is great at routine work but stumbles with complex systems and real-world security demands.
CloudBees’ Chief Product Officer Shawn Ahmed says developers still spend half their time on repetitive tasks, like triaging tests and waiting for builds, leaving only 20–30% of time for writing code. That’s inefficient, like tolerating a manufacturing line running at 20–30% capacity. 
Their solution? Use AI as a companion, not a replacement. Their AI-powered continuous integration tool scans past deployments to suggest configurations, cutting hours from setup. AI handles grunt work—pattern recognition, basic security spotting—while humans decide the hard calls.
Ahmed breaks AI adoption down into three outcomes:
- Replacers try to cut jobs with automation but get poor results. 39.2% of DORA survey respondents don’t trust AI-generated code.
- Tools First! teams grab AI tools without strategy. Gains in some areas come with losses in quality and security.
- Augmenters treat AI as a human boost. 25% more AI adoption links to a 2.1% bump in productivity and better job satisfaction.
CloudBees sticks with augmenting. Their DevSecOps AI flags vulnerabilities but leaves fixes to humans. Their Value Stream Analytics uses AI to spot inefficiencies but lets teams decide next steps.
Five rules from CloudBees to keep AI a hero, not a headache:
- Attack high-toil, repetitive work first.
- Set governance before deploying AI wide.
- Train developers on AI’s strengths and limits.
- Integrate AI deeply, don’t bolt on isolated tools.
- Track quality, security, and business impact, not just code output.
Shawn Ahmed on AI’s future role:
AI is your enthusiastic partner, not your replacement.
AI is adept at the mundane, repetitive tasks that drain motivation and attention among developers and increase their toil. It lets humans focus on innovation, customer experience, and ultimately, business value.
He expects developers soon will have multiple personalized AI companions tailored to their workflow.
Ahmed also warns against using AI just to shrink teams. Instead, it should help teams scale up performance beyond what more people alone could achieve.
The takeaway? AI alone isn’t the hero of software development. Neither is pure human skill. The win comes from pairing them effectively.
