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, based on his own work records. His account describes faster output alongside new pressures, including context switching, weaker shared understanding and concerns about AI usage limits.
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, describing a sharp change in his work after moving to an environment built around AI coding agents. His account, published by The Pragmatic Engineer as the winner of its essay challenge, offers one engineer’s comparison—not a controlled study—of how AI tools and workplace practices can affect software delivery.
Kakkar says he joined Tano as its first engineer and worked there for eight months, building an affiliate platform, improving a content-monitoring system and carrying out a migration of the company’s agent architecture. He compares that work with four years at PostHog, where he worked across product areas including analytics, experimentation, surveys and feature flags. He says the work at Tano covered a similar breadth of product in a much shorter period.
In his figures, Kakkar reports shipping 11 times more pull requests per month at Tano than at PostHog. He says his PostHog rate was 2.4 times his rate at Bloomberg, where he began his career. The source does not provide the underlying monthly counts in the supplied material, and Kakkar cautions that pull-request totals can be gamed. He says the figures are an honest account of his own work, but they do not measure code quality or establish what other engineers can expect.
Kakkar contrasts the tools and review practices at the companies. At PostHog, he says, AI use before late 2024 largely meant tab completion and help with repetitive tasks. At Tano, he describes working with coding agents capable of taking on larger feature-building tasks. He also says previous employers required peer review and had a high code-quality bar; at Tano, reduced communications and review overhead helped him move faster, with a tradeoff of greater risk of breaking things.
Faster Shipping, New Workflow Costs
Kakkar’s account matters because it links AI-assisted output to changes in how engineering teams coordinate and check work, not just to faster code generation. His reported increase suggests that AI tools can change an individual engineer’s throughput when paired with a culture that permits rapid iteration. But the comparison reflects one person’s experience across different companies and stages of his career; it cannot isolate the effect of AI from differences in team size, role, project scope or startup culture.
The essay also describes costs that can accompany faster delivery. Kakkar says the new pace brought more context switching, difficulty maintaining shared context, less focus on design and concern about AI plan limits. These pressures may affect how teams review changes and preserve knowledge as code is produced more quickly. His experience frames the practical question for software organizations as not only how much work agents can accelerate, but how teams maintain quality, focus and coordination.
Products Worth Considering
Bose SoundLink Flex Bluetooth Speaker (2nd Gen) – Portable Outdoor Speaker with Hi‑Fi Audio, Waterproof and Dustproof, USB‑C, Up to 12 Hours
Samsung Galaxy Tab A7 Lite 8.7 32GB Wi‑Fi Android Tablet – Long‑Lasting Battery, Compact Slim Design, Sturdy Metal Frame (US Version)
DJI Osmo Mobile 8 Standard Combo Phone Gimbal Stabilizer
As an affiliate, we earn on qualifying purchases.
From Tab Completion to Agents
Kakkar’s comparison spans three workplaces with different sizes and processes. He describes Bloomberg as a large financial technology company with about 20,000 employees and defined processes. PostHog was a much smaller startup when he joined, growing from around 10 people to about 200, according to his account. Tano had around 10 people when he wrote the essay, and he joined as its founding engineer.
He presents PostHog through late 2024 as a pre-AI baseline, while noting that the company has since adopted AI more extensively and that output is increasing, a statement he makes based on following its progress. The source material does not include independent measurements of that increase. Kakkar says the tools at Tano included Claude Code Desktop with agent support, while at PostHog the main AI assistance he recalls was code completion. The essay was selected from 125 submissions to The Pragmatic Engineer’s challenge marking its fifth anniversary; the publication said it would feature runner-up essays in a later issue.
“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 figures are Kakkar’s personal data, and the supplied report does not include raw monthly pull-request counts, a measurement method, or an independent audit. Pull-request volume alone does not show the size, quality, reliability or user impact of the work. The comparison also spans different roles, companies, team sizes and periods, making it impossible to attribute the full difference to AI agents.
It is also unclear how broadly Kakkar’s experience applies to other engineers or organizations. The essay describes new operational concerns, but provides no quantified measure of time spent on context switching, code defects, design work or AI plan limits. The source material does not establish whether the reported pace can be sustained or what effect it has on long-term maintenance.
Further Essays and Broader Evidence
The Pragmatic Engineer said it planned to publish runner-up essays in a follow-up issue, giving readers additional accounts of changes in tech workplaces and projects. The report does not specify a publication date for that follow-up.
For a clearer picture of AI’s effect on software engineering, readers would need comparable measurements across more engineers and teams, including measures of quality, maintenance and delivery outcomes alongside code volume. Until then, Kakkar’s account is best read as a detailed personal comparison of two working eras, rather than a general estimate of AI’s impact.
Key Questions
Who wrote the comparison?
Neil Kakkar, a London-based software engineer with nine years of professional experience, wrote the essay published by The Pragmatic Engineer.
How much did Kakkar’s pull-request rate change?
He reports merging 11 times as many pull requests per month at Tano as at PostHog, and 2.4 times as many at PostHog as at Bloomberg. These are his personal figures, not independently verified industry statistics.
Does the comparison prove AI caused the increase?
No. Kakkar links the change to AI tools and engineering culture, but the comparison involves different employers, roles and periods. It cannot isolate AI as the sole cause.
What downsides did he describe?
Kakkar cited context switching, loss of shared context, less attention to design and concern about AI plan limits. He also described a tradeoff between reduced review overhead and a greater risk of breaking things.
Source: newsletter.pragmaticengineer.com
