AI helps you write code 50% faster, then the code sits in the review queue for three days.
Nicole Forsgren is the creator of the DORA metrics and the lead author of “Accelerate.” For the past decade, the software industry’s way of measuring “delivery speed” has been largely tied to her research. In November 2025, her new book “Frictionless,” co-authored with DX CEO Abi Noda, was published, directly addressing a question: AI has clearly accelerated everything, so why aren’t teams delivering faster?
In February 2026, Forsgren participated in a conversation at The Pragmatic Summit hosted by Gergely Orosz, covering bottleneck shifts in the AI era, organizational process debt, measurement pitfalls, and change management.
Key Takeaways:
-
• AI coding tools accelerate developers’ “inner loop” (writing code, debugging), but the “outer loop”—code review, CI/CD, deployment—remains unchanged, limiting overall delivery speed improvements. -
• Increased code output without improved review processes is like pouring more water into a clogged pipe. -
• Organizational process debt (meetings, approvals, legacy processes that no longer add value) becomes glaringly obvious in the AI era. -
• 100% AI tool adoption with 0% delivery improvement is entirely possible. -
• Forsgren’s advice for CTOs: Track Developer Flow and reduce the frequency of developer interruptions.
Gergely’s opening question: What has changed and what hasn’t when leading engineering teams in the AI era?
Forsgren’s answer: The fundamentals haven’t changed. Clear goals, short feedback loops, a culture of psychological safety—these elements of high-performing teams remain crucial even with AI. What has changed is the location of bottlenecks.
She used a framework to explain: the inner loop and outer loop.
The inner loop is the developer’s individual work: writing code, running local tests, debugging. Tools like Copilot excel here, genuinely accelerating individual coding efficiency.
But what happens after the code is written? It enters a review queue waiting for someone to look at it, goes through the CI/CD pipeline (automated build and test systems), undergoes security scans, and finally gets deployed to production. This is the outer loop.
We’re making the fast parts faster while the slow parts stay just as slow.
(“We’re making the fast parts faster while the slow parts stay just as slow.”)
Forsgren gave an example: AI helps you write code 50% faster, but code review still takes three days due to team silos or process friction. The overall “value delivery time” hasn’t improved much.
Note: DORA metrics are a software delivery performance framework proposed by the DevOps research project led by Forsgren. It includes four core metrics: deployment frequency, lead time for changes (cycle time from commit to deployment), time to restore service, and change failure rate. The 2018 research findings were compiled into the book “Accelerate.” This set of metrics has been adopted by a large number of engineering teams worldwide.
A Flood of Code Hits the Review Gate
Gergely followed up: Could there be a “code tsunami” hitting those manual review gates?
Forsgren said this isn’t a question of “could,” it’s already happening.
Code output has increased, but review and release processes haven’t kept up, resulting in massive backlogs. Her suggestion: AI shouldn’t just be used to write code; it should be used to review code. Automated testing also needs to be upgraded simultaneously to reduce reliance on manual QA.
If you don’t scale the ‘checking’ with the ‘creating,’ you’re in trouble.
(“If you don’t scale the ‘checking’ with the ‘creating,’ you’re in trouble.”)
When most companies talk about AI coding tools, their focus is entirely on “how to make developers write faster.” Forsgren’s reminder is that writing faster is only half the problem, and perhaps the easier half.
Process Debt Illuminated by AI
The topic shifted to “process debt.” Forsgren introduced a concept: organizational process debt.
All those meetings, approvals, “we’ve always done it this way” steps might have had value when they were created, but they no longer add value now. She believes organizational process debt is one of the biggest killers of high performance.
In the AI era, these legacy processes become more visible and more frustrating. In the past, developers might have silently accepted the fact that “approvals need three levels” because writing the code itself took days anyway. But now AI helps you finish the code in half a day, and then you find yourself spending even more time waiting for various processes to complete. The sense of disparity is particularly strong.
Note: Technical debt refers to expedient solutions taken in software development to meet deadlines, requiring more cost to fix later. The “process debt” proposed by Forsgren is a similar concept but points not to code, but to organizational processes: approval, meeting, and sign-off steps established for some reason long ago, which have long lost their purpose but no one has cleaned them up.