TL;DR
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
Software engineer Neil Kakkar reports merging 11 times as many pull requests per month at AI-native startup Tano as he did at PostHog, where he worked before today’s coding agents became widespread. His account, published by The Pragmatic Engineer, is an individual comparison rather than a controlled study, and it describes new costs alongside higher output, including context switching and weaker shared understanding.
Software engineer Neil Kakkar says he merged 11 times as many pull requests per month at AI-native startup Tano as he did at PostHog, offering a personal account of how coding agents and engineering culture are changing software work. The comparison, published as the winning essay in The Pragmatic Engineer’s writing challenge, records higher output alongside new workflow strains, but does not establish that AI alone caused the increase.
Kakkar writes that before joining Tano he typically merged eight to 20 pull requests a month at PostHog. At Tano, where he became the first engineer, he reports averaging more than 200 a month over his first eight months. The publication summarizes the difference as 11 times more monthly pull requests. He also reports that his PostHog rate was 2.4 times his earlier rate at Bloomberg, where he began his career.
The figures cover work of different durations and company settings, and pull-request counts measure merged code changes rather than software quality or user impact. Kakkar acknowledges that PR counts can be gamed, and says he did not try to manipulate his numbers. He describes working on products across both startups, including PostHog’s analytics and experimentation tools and Tano’s affiliate platform, content monitoring system and agent architecture.
He links the change partly to AI tools: at PostHog through late 2024, he says AI use was largely tab completion and help with repetitive tasks. At Tano, he encountered coding agents capable of doing larger pieces of work. He also credits an engineering culture built around experimentation and fast delivery. The essay presents his own experience, not a measured company-wide productivity study.
More Code, Different Work Pressures
Kakkar’s comparison points to a change in the unit and pace of engineering work: an engineer can now delegate more implementation to AI agents and review or direct a larger stream of changes. For companies, that may alter how teams set expectations, structure reviews and judge individual output. The reported PR increase, however, cannot by itself show that teams deliver better products or that the same gains apply elsewhere.
The essay also describes costs that may matter as much as throughput. Kakkar says faster work brought more context switching, weaker shared context and less time for design, as well as concern about limits on AI plans. If these pressures are common, organizations may need to adapt coordination and quality practices rather than simply expect more code from each engineer. The account identifies possible tradeoffs; it does not quantify their effects.
Products Worth Considering
Logitech MX Mechanical Wireless Illuminated Keyboard, Clicky
Samsung 34-Inch Odyssey G5 Ultra-Wide Curved Gaming Monitor
LG UltraGear 32GX850A-B 32-inch 4K UHD Glossy OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
From Tab Completion to Agents
The comparison spans three distinct workplaces. Kakkar spent three years at Bloomberg, a large financial technology company with established processes, then four years at PostHog, a startup that grew from about 10 people to around 200 during his time there. He joined Tano as its first engineer; the company had about 10 people when he wrote the essay.
He treats his PostHog period as a pre-AI benchmark, noting that modern coding agents were not part of his workflow there. That framing is specific to his own experience: Kakkar says PostHog has since adopted AI more actively and that output is rising, but the source does not provide comparative data for that later period. The Pragmatic Engineer published his essay after receiving 125 submissions to a challenge marking the publication’s fifth anniversary.
“I’m shipping 11x more PRs at Tano than at PostHog, where I shipped 2.4x as many as I did at Bloomberg.”
— Neil Kakkar, software engineer and essay author
What the PR Count Cannot Show
The essay does not provide an independent audit of Kakkar’s figures, a common measurement method across companies or a quality-adjusted measure of output. It also does not isolate the effects of AI tools from company size, role, product scope and engineering culture. The 11-fold figure is his reported comparison, not a general estimate for software engineers.
It remains unclear how much of the higher PR rate translated into faster customer outcomes, fewer defects or lasting maintenance costs. The source gives no controlled comparison between AI-assisted and non-AI work at the same company under otherwise similar conditions. Kakkar’s account also does not establish whether the context-switching and design concerns he describes are widespread across AI-native teams.
Further Essays and Broader Evidence
The Pragmatic Engineer said it would publish a follow-up highlighting runner-up essays and linking to other submissions. Those accounts may offer additional perspectives, but they would still need consistent measures to establish how common the reported changes are.
The next useful evidence would compare work quality, delivery time, defects and maintenance demands alongside code volume, across teams and over time. For now, Kakkar’s report documents one engineer’s experience across two eras: a sharp rise in his pull-request count and a set of new coordination and focus challenges whose scale remains unmeasured.
Key Questions
What did Neil Kakkar report?
He said he merged 11 times as many pull requests per month at Tano as at PostHog. He describes the figure as a personal comparison, not an industry-wide result.
Does the essay prove AI caused the increase?
No. Kakkar connects the increase to AI agents and engineering culture, but his comparison does not separate AI’s effect from differences in role, company, product work or time period.
How does Kakkar define the earlier period?
He treats his work at PostHog through late 2024 as a pre-AI baseline, saying AI there was mainly used for tab completion and repetitive tasks. That describes his experience, not every employee’s tools or current practice at PostHog.
What drawbacks did he describe?
Kakkar reported more context switching, reduced shared context and less focus on design, along with concerns about AI plan limits. The essay does not measure how often those problems occur or their effect on product quality.
Source: newsletter.pragmaticengineer.com
