Tag: engineering
10,000 PRs a month is easy: How devex is evolving at PostHog
Shipping cadence is accelerating at PostHog. In the last 6 months, we've gone from shipping 1,441 PRs in January to 4,725 PRs in June while increasing engineering headcount by only 10%. This is heading towards our north star for internal developer…
108 PRs in eight days: Accidentally discovering loop engineering
Don't build your own CRM
You really don't want to be in the business of building your own Salesforce.
The third horizon of human-centred approaches
The way we work as human-centred practitioners needs to change. From its beginnings as an engineering technique, through its transition to being a product design tool, user experience now needs to be adopted as a strategic approach.
Why Cascade Chat kept disconnecting
Cascade Chat could connect to Libera.Chat just fine and then lose the socket a few seconds later. Finding out why took several PRs and a rethink of its event architecture.
Our session replays are opening their own pull requests
Errors are the easiest data to act on in our product improvement pipeline . They arrive with a stack trace that's groupable and says exactly what broke and where. This makes it easy for agents to research each one, and if it's actionable, open a PR…
How to build a PostHog integration with the provisioning API
I built a fake farm-website company on PostHog's provisioning API. Here's how it creates accounts for its users and reads their analytics back, with the gotchas I hit.
Your Agent Deserves Logs
How structured logs helped Codex fix a year-old bug, and why logs are the key to unlocking autonomous workflows.
Cascade Chat has moved to an ADLC
Cascade Chat now runs on an Agentic Development Lifecycle. Agents write the code, Playwright verifies it, and the only human decisions left are ideation, architecture, and the release gate.
I wrote a 70x faster SQL parser while barely looking at the code
After the success of using agents to improve query performance through autoresearch , I wanted to try something more ambitious. I rewrote PostHog's SQL parser using multiple long-running Claude Code sessions in parallel. The result was 16K lines of…
CalVer is just better
Semantic versioning signals intent. Calendar versioning tells you when. The second one is more useful.
Are we (re-)approaching the year of applications on the desktop?
Desktop software is having a quiet revival. After years of SaaS monoculture, frameworks like Wails and Tauri are enabling a new generation of native-feeling apps built with familiar web technologies — at a fraction of Electron's resource cost. Wails v3's first-class in-app updates might be what finally tips the scale.
I dumped Cursor
As is tradition, my workflow continues to evolve at a rapid pace. I've since dumped Cursor in favor of a new stack.
Treating LLMs as programming books
Thoughts on an approach for using LLMs effectively for coding without losing engagement and cognitive effort.
Treating LLMs as programming books
Thoughts on an approach for using LLMs effectively for coding without losing engagement and cognitive effort.
Karpathy's Autoresearch found a 3-year-old bug in our query engine (and improved performance by 11%)
A few weeks ago at a team offsite in Lisbon, we pointed an AI agent at our query engine, fed it slow queries from production, and let it run overnight. By the next morning it had found something embarrassing: for almost three years, every query with…
Being A 1.5-10x Developer
Good code begets good results, so let's make a case for using coding agents to write better code, rather than just more code.
From 270GB RAM to 5GB: Moving local flag evaluation from Django to Rust
How we moved PostHog's feature flags local evaluation endpoint from Django to Rust and dropped CPU usage by 24x and memory by 56x.
The Teaching Computer
Many people are using AI to solve their problems when they could be leveraging AI to become better versions of who they want to be.
A Guide to Harness Engineering: Building Reliable AI Agent Workflows
A practical framework for building reliable AI agent workflows through harness design, state management, and end-to-end verification.
4,063 errors closed without a human opening PostHog – here's what we learned
Last month, our customers deployed AI agents to PostHog projects to try and solve 6,124 errors in their products. They resolved 4,063 issues. They suppressed 1,751 more. They routed 310 to the right team. Almost none of those agents opened the…
AI Engineer Unconference Sydney 2026
Notes on the AI Engineer Unconference in Sydney around agentic engineering, AI safety, and knowledge management.
Deterministic Governance and Proactive Protection for Autonomous Coding Agents
An in-depth analysis of the critical role of deterministic governance and proactive protection mechanisms, such as hooks, in ensuring the safe and reliable operation of autonomous coding agents like Claude Code and GitHub Copilot CLI.
Untangling Tokio and Rayon in production: From 2s latency spikes to 94ms flat
Learn how we discovered a big performance bottleneck in the Feature Flags service and made Tokio and Rayon play nice with each other.
A beginner's guide to testing AI agents
Software testing used to be relatively straightforward. You wrote business logic, defined expected outputs, and verified them with unit, integration, and UI tests. With the same input, your system behaved the same way every time. AI agents changed…
The Claude Code Leak
What the accidental Claude Code source code leak tells us about the real value of code, product market fit, and why integration is what actually makes software great.
I hate the standup bot (so I built an agent to do it for me)
There it is. Staring at me. Sitting there. Goading me. The day is nearly done. I've reviewed PRs, fixed bugs, hopped on quick calls, commanded my agents, and smashed that accept button all day! And now I'm expected to fill in some stupid Slack thread…
Workslop
The bar for shipping quality work hasn‘t lowered just because AI can generate code in minutes — and if you‘re accepting AI output without reviewing it, you‘re part of the problem.
Small Steps For Agent-Friendly Codebases
Updating an open source library taught me a valuable lesson for making your codebase agent-friendly — small steps can have a big payoff.
The impact of connotations in tech (an Indian pov)
An ecosystem evolves with the meaning and context of the words/conventions used by it's participants. We need shared understanding to have meaningful conversations around topics. The engineering & scientific communities over the world have created common parlance where applicable, where does Indian context fit in?
Code Accountability - 2026
How we caught our AI agent embezzling tokens
The PostHog Wizard is an AI agent that integrates PostHog into your code base. It understands your business logic, captures high quality data, and correctly configures PostHog products in a single command. The Wizard adds incredible stakeholder…
Learn to Love the Rebuild
Refactoring isn't a necessary evil -- it is the job. Why the best product teams rebuild constantly, at every scale.
Learn to Love the Rebuild
Refactoring isn't a necessary evil -- it is the job. Why the best product teams rebuild constantly, at every scale.
Engineering Excellence in the Agentic Era: A Framework for Professional Standards and Quality Control
Engineering Standards and Quality Control in the Age of AI Agents
Automating Codex
You‘re one step away from turning Codex into a proactive assistant that tackles low-hanging fruit while you sleep.
Supporting Markdown Search For LLMs
Learn a simple but effective technique that makes your blog more easily searchable for LLMs.
Updating Your Website For Generative Engine Optimization
Learn how to optimize your website for generative search engines like ChatGPT, Claude, and Gemini to improve your search engine rankings.
MCP Apps And Interactive UIs In MCP Clients
MCP Apps adds an official extension for rendering interactive UIs inside MCP clients, building on MCP-UI and OpenAI's Apps work to move beyond plain text.
Programmatically Setting GitHub Issue Types
A quick guide on how to programmatically set the GitHub issue type with the GitHub CLI, even if built-in functionality for this doesn't quite exist.
Hello, Anthropic
After three years at Microsoft working on security and MCP, I'm joining Anthropic as a Member of Technical Staff to help grow the MCP ecosystem.
Wrapping Up My Latest Microsoft Chapter
Closing out a third stint at Microsoft - reflections on three and a quarter years working on developer experience before moving to a new role in 2026.