Gergely Orosz · Nov 18, 2026
The Pragmatic Engineer: Harjot Gill on software factories and code reviews
Markdown · JSON · Original source
Gergely Orosz of describes a conversation with Harjot Gill, CodeRabbit’s founder and CEO, about building software with . Gill says he shipped over 5,000 pull requests in a month, often using agents for maintenance work such as fixing flaky tests and improving CI. Orosz presents CodeRabbit’s heavy AI spending as potentially justified when it helps the company create products it can sell, but says this is not a typical case for companies whose existing customers may not pay more for new development. Gill argues that agents need ways to verify their own work. The methods he describes include formal models, simulations, emulated and staged production rollouts. He expects engineers to spend less reviewing individual agent-written pull requests and more time reviewing plans and releases. At CodeRabbit, Orosz says, engineers review releases made up of multiple pull requests. The account describes a shift in human judgment, not its disappearance: review moves toward planning and releases, while the team monitors rollouts and holds people accountable when releases cause outages. Orosz links CodeRabbit’s high AI spending to its business and product development, rather than presenting it as a general template. He also notes Gill’s report of disturbed sleep from waking at night to nudge agents.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Press play to watch. If playback is unavailable, open the source below.
Original source
· Article
Source details
Original source date: Nov 18, 2026 (2026-11-18T18:08:52+00:00).
Date basis: Date stated by the publisher.
Added to Escape Meridian: 2026-11-18T18:17:32.041Z. First observed: 2026-11-18T18:13:20.855Z. Retrieved: 2026-11-18T18:16:32.639Z.
Referenced by
Recorded citations in public readings. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
The account reports Gill’s and Orosz’s descriptions; it does not provide independent measurement of the reliability or quality of the software shipped. The source gives examples of verification methods but no comparative results showing how well they prevent production failures. The page summarizes a podcast conversation and links to its transcript; this reading relies on the supplied page text.
Open question
What evidence from CodeRabbit’s experience would show whether its verification methods keep agent-built software reliable as its output grows?
Source-stated
Orosz reports that Gill shipped more than 5,000 pull requests in a month, using agents for maintenance work as well as product development.
Janitorial work formed a large part of his 5,000 PRs last month, pushed with agents.
Source-stated
Orosz frames CodeRabbit’s AI spending as potentially justified when it helps the company build new products it can commercialize, while cautioning that CodeRabbit is not typical.
massive AI spend can be justified by launching new, AI-related products faster
Source-stated
Gill says agents need ways to verify their own work, and the account describes methods from formal models and simulations to staged rollouts.
Harjot believes agents need to be “hill climbing”; they need to have ways to verify their own work
Source-stated
Gill expects human review to shift from individual agent-generated pull requests toward planning and final releases; Orosz says CodeRabbit engineers review releases comprising multiple PRs.
Inside of CodeRabbit, engineers are reviewing releases (multiple PRs), not individual PRs.
Source-stated
Orosz says Gill reports disturbed sleep from waking at night to nudge agents.
Harjot also reports disturbed sleep, like waking up at night to nudge his agents.
Source and contributor context
Gergely Orosz · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
Media declared by the original publisher; availability checked separately from the editorial reading.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
AI agent
An AI system that chooses and carries out actions toward a goal, often using software tools.
API
An application programming interface: a defined way for software to request data or actions from another system.
The Pragmatic Engineer
The Pragmatic Engineer is Gergely Orosz's newsletter on software engineering practice, careers and the tech job market.
TIME
TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Editorial provenance
Version 1. Publication ID: ae379562-4a0b-45fc-87c3-9fb320fad3d5.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-18T18:17:32.041Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: useful_perspective.
Observation: 30da4b3a89438222f31487a7db5c96281a83bfd010e3a42299d9b01866c37016. Source text hash: 7951cbb9db08de87d25a6894bd333078395e209ed9df0fa7b41974d0744695fb.
Revision and source-check history
Version 1 · 2026-11-18T18:17:32.041Z
The Pragmatic Engineer: Harjot Gill on software factories and code reviews
Gergely Orosz of describes a conversation with Harjot Gill, CodeRabbit’s founder and CEO, about building software with . Gill says he shipped over 5,000 pull requests in a month, often using agents for maintenance work such as fixing flaky tests and improving CI. Orosz presents CodeRabbit’s heavy AI spending as potentially justified when it helps the company create products it can sell, but says this is not a typical case for companies whose existing customers may not pay more for new development. Gill argues that agents need ways to verify their own work. The methods he describes include formal models, simulations, emulated and staged production rollouts. He expects engineers to spend less reviewing individual agent-written pull requests and more time reviewing plans and releases. At CodeRabbit, Orosz says, engineers review releases made up of multiple pull requests. The account describes a shift in human judgment, not its disappearance: review moves toward planning and releases, while the team monitors rollouts and holds people accountable when releases cause outages. Orosz links CodeRabbit’s high AI spending to its business and product development, rather than presenting it as a general template. He also notes Gill’s report of disturbed sleep from waking at night to nudge agents.
Original source
· Article
Source details
Original source date: Nov 18, 2026 (2026-11-18T18:08:52+00:00).
Date basis: Date stated by the publisher.
Added to Escape Meridian: 2026-11-18T18:17:32.041Z. First observed: 2026-11-18T18:13:20.855Z. Retrieved: 2026-11-18T18:16:32.639Z.
Limits
The account reports Gill’s and Orosz’s descriptions; it does not provide independent measurement of the reliability or quality of the software shipped. The source gives examples of verification methods but no comparative results showing how well they prevent production failures. The page summarizes a podcast conversation and links to its transcript; this reading relies on the supplied page text.
Open question
What evidence from CodeRabbit’s experience would show whether its verification methods keep agent-built software reliable as its output grows?
Source-stated
Orosz reports that Gill shipped more than 5,000 pull requests in a month, using agents for maintenance work as well as product development.
Janitorial work formed a large part of his 5,000 PRs last month, pushed with agents.
Source-stated
Orosz frames CodeRabbit’s AI spending as potentially justified when it helps the company build new products it can commercialize, while cautioning that CodeRabbit is not typical.
massive AI spend can be justified by launching new, AI-related products faster
Source-stated
Gill says agents need ways to verify their own work, and the account describes methods from formal models and simulations to staged rollouts.
Harjot believes agents need to be “hill climbing”; they need to have ways to verify their own work
Source-stated
Gill expects human review to shift from individual agent-generated pull requests toward planning and final releases; Orosz says CodeRabbit engineers review releases comprising multiple PRs.
Inside of CodeRabbit, engineers are reviewing releases (multiple PRs), not individual PRs.
Source-stated
Orosz says Gill reports disturbed sleep from waking at night to nudge agents.
Harjot also reports disturbed sleep, like waking up at night to nudge his agents.
Source and contributor context
Gergely Orosz · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
Media declared by the original publisher; availability checked separately from the editorial reading.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
AI agent
An AI system that chooses and carries out actions toward a goal, often using software tools.
API
An application programming interface: a defined way for software to request data or actions from another system.
The Pragmatic Engineer
The Pragmatic Engineer is Gergely Orosz's newsletter on software engineering practice, careers and the tech job market.
TIME
TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Editorial provenance
Version 1. Publication ID: ae379562-4a0b-45fc-87c3-9fb320fad3d5.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-18T18:17:32.041Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: useful_perspective.
Observation: 30da4b3a89438222f31487a7db5c96281a83bfd010e3a42299d9b01866c37016. Source text hash: 7951cbb9db08de87d25a6894bd333078395e209ed9df0fa7b41974d0744695fb.
uncertain · 2026-11-18T18:17:32.041Z · anthropic/claude-opus-5.5
Publication: ae379562-4a0b-45fc-87c3-9fb320fad3d5
attribution — pass: The draft attributes the framing and observations to Gergely Orosz and the claims about PR counts, verification and review to Harjot Gill, CodeRabbit's founder and CEO. This matches the source.
Harjot Gill is founder and CEO of CodeRabbit, an AI code review business
summary — uncertain: The draft blunts Gill's central thesis. The source says he believes human code review is dead and that 'we will stop reviewing code generated by agents.' The draft says engineers will 'spend less time reviewing individual' PRs. Smallest repair: state that Gill argues human review of agent code is dead and will stop, with review moving to plans and releases. The draft also omits the $250,000-a-month personal AI spend that frames the episode. Central terms such as pull request, CI and staged rollout are not explained and need term tooltips.
Harjot thinks that we will stop reviewing code generated by agents.
date — pass: The originalPublished date, 2026-11-18, matches the page's 'Nov 18, 2026' date and the feed entry.
Gergely Orosz Nov 18, 2026
support — uncertain: The claims and excerpts are source-grounded. One synthesis line says 'the team monitors rollouts.' The source says an agent often observes the rollout, checks logs and continues it, so this shifts the role from the agent to humans. Smallest repair: say an agent often monitors rollouts. The accountability point, the strike system tied to outages, is supported. The question and limitations are appropriately non-leading.
He said they often have an agent observe the performance characteristics of the rollout, check the logs, and continue the wider rollout when things are stable.
media — pass: The selected media is the verified original article page, which matches the source and gives direct access to the episode and transcript. The verified YouTube video would also have been relevant.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
The Pragmatic Engineer is Gergely Orosz's newsletter on software engineering practice, careers and the tech job market.
Explore The Pragmatic EngineerAn AI system that chooses and carries out actions toward a goal, often using software tools.
Definition referenceAn application programming interface: a defined way for software to request data or actions from another system.
Definition referenceTIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Explore TIMEThe Pragmatic Engineer is Gergely Orosz's newsletter on software engineering practice, careers and the tech job market.
Explore The Pragmatic EngineerThe Pragmatic Engineer is Gergely Orosz's newsletter on software engineering practice, careers and the tech job market.
Explore The Pragmatic EngineerAn AI system that chooses and carries out actions toward a goal, often using software tools.
Definition referenceAn application programming interface: a defined way for software to request data or actions from another system.
Definition referenceTIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Explore TIMEThe Pragmatic Engineer is Gergely Orosz's newsletter on software engineering practice, careers and the tech job market.
Explore The Pragmatic Engineer