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 more pull requests per month at AI-native startup Tano than at PostHog, where his work preceded widespread coding agents. His account, published by The Pragmatic Engineer as the winning essay in a writing challenge, describes faster shipping alongside more context switching and concerns about design and AI plan limits.
Software engineer Neil Kakkar says he merged 11 times more pull requests per month after moving from PostHog to AI-native startup Tano, in a first-person account published by The Pragmatic Engineer. The comparison offers a detailed but individual snapshot of how coding agents may change engineering output—and the work pressures that can accompany faster delivery.
Kakkar reports that he shipped about 11 times as many pull requests at Tano as at PostHog, where he worked for four years. He says his PostHog output was itself around 2.4 times his output at Bloomberg, where he began his career. The report describes the Tano figure as more than 200 pull requests a month, compared with a prior range of eight to 20 per month before the move. These are his own work records, not an industry-wide measurement.
The comparison spans different companies, roles and periods. Kakkar joined Tano as its first engineer and spent eight months there building an affiliate platform, improving a content-monitoring system and migrating the company’s agent architecture. At PostHog, over four years, he worked across analytics infrastructure and product development, including experimentation, surveys and early-access features. He says AI tools were largely limited to tab completion at PostHog through late 2024, while agent-based coding tools were available when he joined Tano.
Kakkar warns that pull-request counts can be gamed and says he did not set out to inflate his numbers. He attributes the increase to both AI and the engineering culture that enabled faster work. His account also describes costs: more context switching, less shared understanding, reduced attention to design and concerns about limits on AI plans. The publication says the essay won a challenge that drew 125 submissions and carries a $10,000 grand prize.
More Output, New Workflow Costs
Kakkar’s account illustrates why AI’s effect on software engineering cannot be captured by code-generation speed alone. If agents let an engineer complete more implementation work, teams may also need to reconsider how they coordinate, review changes and preserve shared context. The essay suggests that removing old bottlenecks can expose new ones, including the attention required to manage multiple tasks and check what tools produce.
The reported increase may interest developers and employers weighing AI adoption, but it does not establish that similar results are typical. Kakkar’s work volume is one person’s experience across startups with different products, responsibilities and team practices. His account is useful as a concrete case, not a controlled test of AI productivity or evidence that pull-request volume alone measures quality.
Products Worth Considering
DJI Osmo Mobile 7P Phone Gimbal with Mic Mini Transmitter, Infinity Black
LG UltraGear 34GX90SA-W 34-inch WQHD Curved OLED Gaming Monitor
Apple Watch Series 12 GPS 46mm Smartwatch
As an affiliate, we earn on qualifying purchases.
Three Companies, Two Eras
Kakkar’s comparison runs across Bloomberg, PostHog and Tano, which he describes as substantially different workplaces. Bloomberg was a large financial technology company with defined processes; PostHog grew from a small startup into a company of about 200 people during his time there; Tano had around 10 people when he wrote. He says peer review was a high bar at Bloomberg and PostHog, and that seniority brought substantial time spent reviewing code.
The essay treats PostHog as a pre-AI baseline, not as a description of its current practices. Kakkar says AI use at his former company has since expanded and that its output is increasing, while noting he did not use those later developments in his historical comparison. At Tano, he describes working with coding agents capable of handling larger tasks than the autocomplete tools he had used previously.
“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 Output Figures Cannot Show
The source report does not provide an independent audit of the pull-request records, a shared definition of a pull request across companies, or a measure of the quality and impact of the code. The comparison also cannot isolate AI’s effect from differences in company size, role, product scope, experience and engineering culture. Kakkar himself says PostHog’s figures represent a pre-AI period and that the company’s approach has since changed.
It is also unclear how much of the reported speed can be sustained as Tano grows, or how the added work affects defects, maintenance and outcomes for users. The essay describes concerns about attention, design and AI plan limits, but does not quantify those effects or compare them with the time saved.
Further Essays and Team Changes
The Pragmatic Engineer said it would publish a follow-up highlighting runner-up essays and linking to other submissions. For engineering teams, the practical questions raised by Kakkar’s account remain open: whether higher output holds over time, what review and coordination practices can keep pace, and how to judge results beyond the number of pull requests merged. The source report does not announce a formal study or a next measurement of Kakkar’s work.
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 and joined Tano as its first engineer.
What does the 11-times figure measure?
It is Kakkar’s comparison of the pull requests he merged per month at Tano and PostHog. It is a personal output measure, not a controlled estimate of AI’s effect across the software industry.
Does the essay prove AI makes engineers 11 times more productive?
No. It reports one engineer’s experience across different companies and roles. The account does not isolate AI from other factors or measure code quality, defects or user impact.
What difficulties did Kakkar report?
He described context switching, lost shared context and less focus on design, as well as concerns about AI plan limits. The report does not quantify these costs.
Source: newsletter.pragmaticengineer.com
