So we built a system that fixes the allocation. It’s called Diff Vader: an AI reviewer that looks at every PR across our orgs and, for a carefully bounded set of low-risk changes, becomes the only reviewer needed, freeing human reviewers to spend their judgment where judgment actually matters.
Dave NorthLeads the technical operations group · Leader, 
Rewind
Post: Review is the bottleneck now: How we let AI approve pull requests (safely), Rewind blog, June 1, 2026, at Opening.
Dave North, Leads the technical operations group. Writes as the leader of the group that built and runs the Diff Vader reviewer, and says he reports its results to leadership and auditors.
Checked against the source by a first reader, then confirmed independently by a second reader who fetched the source without seeing the first reader's notes (Independent review agent (Claude Sonnet), blind second pass from passage text and citation only, no repository access, 2026-11-04).
Exact match, ignoring a stray space from the link or bold markup. Author Dave North.
Dave North (Rewind), “Review is the bottleneck now: How we let AI approve pull requests (safely)”, Rewind blog, 2026-06-01, at Opening. https://rewind.com/blog/ai-approve-pull-requests-safely/ Archived at https://the-ai-record.com/passages/rewind-diff-vader-ai-reviewer-only-reviewer-needed-low-risk-2026
Auto-approve is not auto-merge. When Diff Vader approves, it posts an APPROVE review, the same thing a human reviewer clicking “approve” does. It does not push the merge button. The engineer merges (or GitHub’s native auto-merge does, if they enabled it), and every existing gate—CI, branch protection, required-reviewer counts—still applies.
Dave NorthLeads the technical operations group · Leader, Rewind
Post: Review is the bottleneck now: How we let AI approve pull requests (safely), Rewind blog, June 1, 2026, at What an engineer actually sees.
Diff Vader and the Death Star were built almost entirely with Claude Code, but not by typing “build me an AI reviewer” and shipping whatever came back. We used spec-based development — Superpowers for the core methodology, augmented with our own skills — which forces every piece of work through brainstorm → written spec → reviewed implementation plan → execution → code review, with discrete human checkpoints in between.
Dave NorthLeads the technical operations group · Leader, Rewind
Post: Review is the bottleneck now: How we let AI approve pull requests (safely), Rewind blog, June 1, 2026, at What I’d tell another engineering leader.
It declares the hard rules the verdict engine is not allowed to override no matter what the council says: paths that touch authentication, authorization, secrets, encryption, and anything that changes who is allowed to approve future changes always require a human.
Dave NorthLeads the technical operations group · Leader, Rewind
Post: Review is the bottleneck now: How we let AI approve pull requests (safely), Rewind blog, June 1, 2026, at The council: many specialists, one question.
We were merging on the order of forty PRs a day across our GitHub organizations, well over a thousand a month, and climbing.
Dave NorthLeads the technical operations group · Leader, Rewind
Post: Review is the bottleneck now: How we let AI approve pull requests (safely), Rewind blog, June 1, 2026, at Opening.
For the curious: reviews average around $0.73 per PR, with significant variance. Small dependency bumps cost cents; large multi-file PRs with many substantive findings can push past twelve dollars.
Dave NorthLeads the technical operations group · Leader, Rewind
Post: Review is the bottleneck now: How we let AI approve pull requests (safely), Rewind blog, June 1, 2026, at The Death Star.
Something wrong? Report a correction. JSON: /api/v1/passages/rewind-diff-vader-ai-reviewer-only-reviewer-needed-low-risk-2026.json