TL;DR
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
Software engineer Neil Kakkar says he merged 11 times more pull requests per month at AI-native startup Tano than he did at PostHog, where his work preceded today’s coding agents. His account, published as the winning essay in The Pragmatic Engineer’s challenge, describes faster output alongside new pressures, including context switching and weaker shared understanding.
Software engineer Neil Kakkar says he merged 11 times more pull requests per month after joining AI-native startup Tano than he had at PostHog, according to a first-person essay published by The Pragmatic Engineer. His comparison offers an individual account of how coding agents and different engineering practices can change the pace of software work, while also creating new coordination and focus problems.
Kakkar reports that his pull-request output at PostHog was about 2.4 times his earlier level at Bloomberg, and that at Tano it was 11 times his PostHog rate. He describes Bloomberg as a large financial company with defined processes, PostHog as a fast-moving startup, and Tano as a small AI-native company where he was the first engineer. The figures are his own work records, not an independently audited measure of productivity.
He worked at PostHog for four years, contributing to analytics infrastructure and products including experimentation, surveys and feature flags. At Tano, he says, he worked on an affiliate platform, content monitoring and a migration of the company’s agent architecture. He says work of comparable breadth was compressed into eight months at Tano, rather than the 45 months he uses for the PostHog comparison.
The working methods also differed. Kakkar says PostHog’s pre-AI workflow relied on peer review, with AI largely limited to code completion and help with repetitive tasks. At Tano, he describes using coding agents capable of producing features with less direct human input. He credits both AI tools and engineering culture for the change, while noting that pull-request counts can be gamed and do not, by themselves, establish software quality or business value.
Faster Shipping, New Engineering Costs
The account illustrates why AI adoption is becoming an organizational issue as well as a tooling choice. If one engineer can produce and merge code more quickly, teams may be able to test ideas and deliver product changes sooner. But more code also means teams must decide how to maintain review quality, keep colleagues informed and preserve a shared understanding of systems.
Kakkar identifies context switching, loss of shared context and reduced attention to design as problems in the faster workflow. He also raises concern about AI usage limits. These observations are relevant to engineering leaders weighing adoption: higher output counts are a signal of changed activity, not proof that reliability, customer outcomes or total productivity improved by the same amount.
Products Worth Considering
Oral-B iO Series 3 Rechargeable Electric Toothbrush for Adults, Black, 3 Modes, 2 Brush Heads and Travel Case
Levoit Core 600S-P Smart Air Purifier for Large Rooms
Philips Sonicare 6500 Series Rechargeable Electric Toothbrush with 2 Brush Heads, White
As an affiliate, we earn on qualifying purchases.
From Completion Tools to Coding Agents
Kakkar’s comparison spans three workplaces and different career stages: Bloomberg, where he began as an engineer; PostHog, where he spent four years; and Tano, where he joined as a founding engineer. He describes Bloomberg as a company of roughly 20,000 people and says PostHog grew from around 10 people when he joined to about 200. Tano had about 10 people at the time of his account.
His PostHog figures are explicitly a pre-AI baseline, drawn from work before the company’s later adoption of AI. Kakkar says that by late 2024 AI at PostHog was mainly used for tab completion and repetitive work; when he joined Tano, he encountered stronger agent support. He also cautions that PostHog has since adopted AI more extensively, so the comparison is between periods and workplaces, not a controlled test of two tools.
“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, in his essay published by The Pragmatic Engineer
What the Output Figures Cannot Show
The reported 11-fold difference is Kakkar’s individual comparison, not a study of engineers generally. The source provides no independent audit of his pull-request counts, no common measurement method across all three employers, and no controlled comparison separating AI’s effect from differences in company size, job scope, team structure or career experience.
The account also does not quantify whether the faster shipping changed defect rates, maintenance costs, customer satisfaction or business results. Kakkar describes concerns about quality, focus and shared context, but the essay excerpt does not provide measurements of those effects or establish how common they are across other teams.
Further Essays and Wider Evidence
The Pragmatic Engineer said it would publish selected runner-up essays in a follow-up issue. Those accounts may add experiences from other engineers, though they would still be personal reports rather than a systematic assessment of AI’s effect on software work.
For now, the practical next step for readers is to treat Kakkar’s figures as a case study and ask what their own teams measure alongside code volume: review burden, defects, delivery time and whether knowledge remains shared. The source does not announce a broader data study or a specific follow-up on Tano’s results.
Key Questions
What changed in Neil Kakkar’s reported output?
Kakkar says he merged 11 times more pull requests per month at Tano than at PostHog. He also reports PostHog output at about 2.4 times his earlier rate at Bloomberg.
Does the comparison prove AI caused the increase?
No. Kakkar credits AI and engineering culture, but this is a personal comparison across different companies and periods, not a controlled study that isolates AI’s effect.
What drawbacks did Kakkar identify?
He describes more context switching, weaker shared context, less focus on design and concern about AI usage limits. The report does not quantify these effects.
Are the PostHog figures representative of its current engineering practices?
No. Kakkar says the PostHog figures are a pre-AI baseline and that the company has since adopted AI more extensively.
Source: newsletter.pragmaticengineer.com
