Incident management platform provider Rootly has abandoned its long-standing small pull request rule, concluding that the practice no longer makes sense now that AI agents generate most of its code. The company published an account of the decision, describing a shift from measuring PR size to assessing blast radius, with feature flags and rollback capability taking precedence over line counts. For two years, Rootly enforced a strict small-PR culture with stacked PRs and atomic changes limited to a few hundred lines, but co-founder and CTO Quentin Rousseau now argues that approach has become obsolete in an AI-driven workflow.
Rousseau explains that the small-PR rule made sense when humans wrote code by hand, as smaller diffs were easier to review and revert. But AI agents have changed the equation because they think in features rather than increments, producing complete implementations—including migrations, models, services, controllers, tests, and frontend components—in a single output. Attempts to make AI agents produce stacked PRs resulted in technically correct but contextually worse outcomes, with review comments on one PR often depending on decisions made in another, forcing reviewers into mental gymnastics across multiple tabs. The company determined that the small-PR rule was optimized for human writing speed, and AI removed that constraint while turning the rule into overhead.
According to the Rootly engineering team, "AI bugs are context bugs. The code works, but it is applied to the wrong thing." They cite examples like a migration that drops a column still in use by a background job, or a service that writes to a table another team reads from. In response, Rootly stopped reviewing AI code the way it reviewed human code and built an internal AI code reviewer that evaluates every PR against engineering standards, producing a structured review with a risk assessment, standardization score, confidence score, and specific findings grouped by severity. The team writes that "the size of the diff stopped being the useful signal. The blast radius is."
The change reflects a broader industry rethinking of pull request workflows at agent speed. At QCon London 2026, Michael Webster discussed how massive AI-generated pull requests create a severe bottleneck for human reviewers and introduce persistent technical debt. At AI Native DevCon London in June 2026, a panel featuring Patrick Debois—now DevRel at Tessl and widely credited as the godfather of DevOps—argued that pull requests make sense in open source, where contributors might not be strategically aligned and trust has to be earned, but inside a team that shares context and goals, the overhead of a PR review cycle is increasingly hard to justify when agents are moving fast. Debois and fellow panelists described how agent costs are forcing process discipline, with measurable token costs making waste literally show up in the bill.
Rootly's philosophy is now to ask the questions that actually predict production incidents. The company's "Why and What" sections force the author to explain the motivation and scope of a change as well as its likely impact. For AI-authored PRs, the human who prompted the agent fills these in—Rootly explicitly instructs AI assistants not to generate these sections, because the whole point is capturing context the AI doesn't have: why this change, why now, what's the business reason. Every PR needs to describe how to safely undo itself, including any data fixes. The use of feature flags has moved the safety boundary from merge to rollout, as every significant feature now ships behind a flag, turned off when the PR has been merged and the code pushed to production. The real review happens during progressive rollout: with the feature enabled for the team first, then a handful of customers, then 10 percent, then everyone. Rousseau closes by explaining that killing the small-PR rule—a process that felt virtuous—felt uncomfortable at first, but was necessary to support the outcome of shipping reliable software quickly. The shift represents a fundamental reordering of software delivery priorities as teams orchestrating AI agents face pressures human-only workflows never had to address. Organizations betting heavily on autonomous code generation will need to decide whether their existing quality gates protect against the failure modes that matter, or simply preserve rituals designed for an earlier era.