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 says he merged 11 times as many pull requests per month at AI-native startup Tano as at PostHog, based on his own work data. His account describes how AI agents and a permissive shipping culture changed his workflow, alongside costs including context switching and reduced time for design.
Software engineer Neil Kakkar says he merged 11 times as many pull requests per month at AI-native startup Tano as he had at PostHog, attributing much of the difference to AI tools and engineering culture. The comparison, published by The Pragmatic Engineer as the winning entry in its essay challenge, is a personal account rather than a controlled study of software teams.
Kakkar reports that his work at Tano compressed a broad range of product development into eight months, compared with 45 months at PostHog. At PostHog, he worked on areas including analytics infrastructure, experimentation, feature flags, surveys and early-access features. At Tano, he says he built an affiliate platform, improved a content-monitoring system and helped migrate the company’s agent architecture.
His earlier employers offer different reference points. Kakkar describes Bloomberg as a large company with defined processes, where he was earlier in his career, and PostHog as a fast-moving startup that grew from about 10 employees to around 200 during his time there. He says he shipped 2.4 times as many monthly PRs at PostHog as at Bloomberg, then 11 times as many at Tano as at PostHog.
Kakkar cautions that pull-request counts can be gamed, but says he did not set out to inflate his numbers. He describes PostHog’s pre-AI period as a baseline in which AI use was mostly code completion and repetitive task assistance. At Tano, he says, agent tools could take on much larger coding tasks. He also identifies costs: context switching, lost shared context, less focus on design and concern about usage limits.
Higher Output, Different Bottlenecks
Kakkar’s account captures a practical question for engineering teams: if AI agents allow an engineer to produce and merge code more quickly, the surrounding review, planning and coordination systems may become constraints. His experience suggests that simply increasing the volume of code is not the whole story; teams also need ways to preserve shared understanding and make design decisions.
The report is a single engineer’s comparison across companies, roles and stages of his career. It does not establish that AI alone caused the reported increase, or that higher PR volume means better software or greater productivity across an organization. Still, the detailed account offers a concrete example of how AI-assisted development can change an individual workflow and bring new pressures along with speed.
Products Worth Considering
Apple Watch SE (2nd Gen) GPS 40mm with Starlight Aluminum Case and Sport Band (S/M)
Amazon Kindle Scribe (32GB) with Premium Pen, Tungsten
LG UltraGear 27GX704A-B 27-inch QHD OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
From Code Completion to Agents
Kakkar spent four years at PostHog and joined Tano as its first engineer, according to the essay. His comparison spans a period when the tools available to him changed: he says that, at PostHog through late 2024, AI was largely limited to tab completion and repetitive coding assistance. At Tano, he encountered agent software capable of handling substantially larger tasks.
The source report says The Pragmatic Engineer received 125 submissions to an essay challenge marking its fifth anniversary and selected Kakkar’s entry for publication as the winner. The newsletter describes his figures as detailed personal work data, while noting that his Bloomberg numbers are rough. It also says PostHog’s AI practices have since evolved, so the pre-AI comparison should not be read as a description of the company today.
“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, in his essay
What the PR Counts Cannot Show
The reported increase is based on one engineer’s output across different companies and career stages. The source does not provide an independent audit, a standardized measure of task difficulty or quality, or company-wide comparisons. Kakkar also notes that his Bloomberg figures are rough. It is not possible from the essay alone to separate the effects of AI tools from changes in role, team size, product scope, company culture or his own experience.
The account does not quantify whether the higher PR count produced better customer outcomes, fewer defects or lower development costs. It also does not establish how common the same results are across other engineering teams. The newsletter says PostHog’s AI use has changed since the pre-AI period covered, but gives no comparable updated figures in the supplied report.
More Essays and Team Evidence
The Pragmatic Engineer said it plans to publish a follow-up highlighting runner-up essays and linking to other submissions. Those accounts may offer additional perspectives, but the source provides no publication date or promise of comparable output data.
For now, Kakkar’s figures should be read as a documented personal experience, not a forecast for software engineering as a whole. Broader conclusions would require comparable measures across teams, including code quality, maintenance, delivery outcomes and the time spent on review and coordination.
Key Questions
Who is Neil Kakkar?
Neil Kakkar is a London-based software engineer with nine years of professional experience, according to The Pragmatic Engineer. He has worked at Bloomberg, PostHog and Tano.
What does the 11-times figure measure?
It refers to Kakkar’s reported monthly pull requests merged at Tano compared with PostHog. It is his personal metric, not a measurement of company-wide output or software quality.
Does the essay prove AI caused the increase?
No. Kakkar attributes much of his increased output to AI and engineering culture, but the comparison is not a controlled study. Differences in role, experience, company and project scope may also matter.
What drawbacks did Kakkar report?
He describes context switching, loss of shared context and less focus on design, as well as difficulty pacing work and concerns about AI usage limits.
Is PostHog still being described as pre-AI?
No. The essay uses Kakkar’s period at PostHog as a pre-AI baseline, but The Pragmatic Engineer says PostHog’s AI use has since changed. The comparison does not describe its current practices.
Source: newsletter.pragmaticengineer.com
