Scale to 1M Users: Simplicity Over Over‑Engineering
If I had to scale an app from 0 to 1 million users, here's what I'd do (step by step): There are different ways to scale & this is just one of them. The best solution & numbers will depend on your needs. Keep it simple and avoid over-engineering. Remember, everything is a tradeoff. What else would you add?
Agents Need User Interaction History, Not Just Code
Banger of a piece, from start to finish. 🔥 "We have handed agents the codebase, the tests, the docs, the specs, the commit history." "We haven't handed them the one artifact that actually survived — the record of what users have been...
IPhone Was a Small Mac; iOS Mirrors macOS
The funny thing about people saying the MacBook Neo is just an iPad or an iPhone is that under the hood iOS is fundamentally the same thing as macOS, and indeed when Apple launched the iPhone they said explicitly...
Solving Real Pain, Community Focus Drives HashiCorp’s Success
9 interesting observations from my conversation with Mitchell Hashimoto (@mitchellh, creator of Ghostty, founder of HashiCorp): 1. Vagrant was created because dev environment setup was an unbillable time sink at a consultancy. At the Ruby on Rails shop where Mitchell worked, jumping...
Spec‑Driven Development Essential for Effective AI Tooling
If your team isn’t discussing Spec-Driven Development or how your design decisions are documented, versioned, and shared, you’re undermining your AI tooling strategy.
AI Makes System Design Impossible to Ignore for Engineers
Most engineers tune out when they hear “system design.” It feels like overhead. Until AI forces you to care about it.
Figma’s Silent Change Signals Fight for Design Infrastructure Control
In February 2026, Figma released version 126.1.2. One line in the changelog caught nobody's attention. It should have. The update stripped --remote-debugging-port on startup, with no announcement and no deprecation warning. Just gone. Overnight, developers who'd spent months building automation...
Master AI Systems, Not Just Tools, to Stay Relevant
If you want to stay relevant in AI by 2026, don’t just learn tools — master systems. Here are 9 AI skills that actually compound: • Multimodal AI (text + image + audio + video) • AI Tool Stacking • LLM Evaluation & Management •...
Quantify Product’s ROI: Board Demands Clear Value
Product is one of the biggest investments in a B2B scale-up — yet its commercial impact is often unclear. When the board asks, “What value did product actually create this quarter?” many teams struggle to answer clearly. 🗓 Wednesday, 4 March ⏰ 17:00 – 18:00...
Structured AI Conversations Mirror Whiteboarding, Align Designs Early
NEW POST @techygarg uses a structured conversation with an AI agent that mirrors whiteboarding with a human: progressive levels of design alignment, reducing cognitive load, and catching misunderstandings at the cheapest possible moment. https://t.co/axw3dnhjhI
From AI Skepticism to New SRE Perspective in One Year
Less than a year ago, Fred and I gave the closing keynote at SRECon25. I can hardly connect with the way I felt back then, or the pitch I made for why skeptical SREs should engage with AI. If I was...
Human‑AI Collaboration Unlocks Exponential Work Transformation
I didn't write this to add to the noise about AI replacing jobs. I wrote it because the real story...human + AI achieving things neither can alone...needs to be the real story. "Work Reimagined" is a blueprint for exponential performance...
AI‑generated Code Speeds Delivery, but Reliability Suffers
On one end, the Anthropic team is a massive user of AI to write code (80%+ of all code deployed is written by Claude Code). They ship amazingly fast. On the other hand, seeing these beyond terrible reliability numbers suggests there...
Observability, Control Flow, Interruption: Key to Safe Agent Orchestration
Terrific thread on agent orchestration architectures. "If an agent started making confident but wrong decisions, how many actions would execute before I could stop it?" The three magic words are "observability", "control flow ownership", and "interruption".
Replace Functionality One‑for‑one Before Adding Features
Migrating systems? Focus on replacing existing functionality one-for-one before adding new features. This phased approach ensures a smoother transition and allows for innovation later. #TechMigration #SystemUpgrade https://t.co/hQN18iMJ1F
Senior Engineers Champion Junior Hiring, Not Execs
Terrific new paper in the ACM from @shanselman and Mark Russinovich, proposing an apprenticeship model for junior engineers. https://t.co/XnQcH3emkj I'll add one thing. At every place that I have seen start hiring junior engineers in the last few years, that charge...
Master System Design with 15 Essential Case Studies
If you want to become good at system design, learn these 15 case studies (save this now):
AI Boosts Engineer Efficiency 100x, Redefining Open‑Source Costs
I cannot stop thinking about the implications that Cloudflare / Vinext has on commercial open source, and in general, the cost of migrations, rewrites, and maintenance. One engineer, with AI, proved to be ~100x as efficient as before. This will have...
AI Makes Open‑source Rewrites Trivial, Cloudflare Proves It
We will see much, much more of this happening. AI is changing open source incredibly rapidly. Rewriting an open source project to a new language/framework used to be a massive effort: AI is making it trivial as Cloudflare just showcased with...
AI Restores Power to Veteran Engineering Leaders
AI saved the non-technical engineering leader. For years, losing hands-on skills felt like decline. Now? Architecture experience plus AI equals leverage. Typing speed doesn’t win anymore. Judgment does. And the people with 20 years of it just became dangerous again.
Mitchell Hashimoto’s Workflow Transformed by AI Tools
How has the day-to-day workflow of Mitchell Hashimoto (@mitchellh) changed, thanks to AI tools? Timestamps: 00:00 Intro 07:19 HashiCorp origins 18:22 The 2010s startup scene in SF 23:11 Funding HashiCorp 25:23 The "Hashi stack" 38:28 The open-core pivot 48:08 Taking HashiCorp public 51:58 The almost-VMware acquisition 59:10 Mitchell’s take...
Prioritize P95/P99 Metrics to Empower Power Users
If only more products would measure p95 / p99 metrics and act on them, instead of looking at medians (p50) or averages (that mask outliers) p99 is almost always your power users. Fixing stuff for them has outsized impact Great example on...
AI's Diverse Uses: From Business to Biological Training
Fragments: how organizations are using AI, reflections from the Utah retreat, agentic engineering patterns, inserting friction for security, training biological neural networks https://t.co/lrzsTVy1gs
AI Overlays Legacy Observability, Slowing Debugging Further
Adding AI to legacy observability practices won't make debugging faster. It'll just amplify the problem.
Priming LLM with Codebase Knowledge Improves AI Workflow
NEW POST @techygarg sees a frustration loop when working with AI and identified five patterns to help. Here's the first: priming the LLM with knowledge about the codebase and preferred coding patterns. https://t.co/yjRPvFChd0
Why Connecting AI Everywhere Fails, Embrace Autonomous Agents
Built an MCP server recently? Did developers actually use it, or is it collecting dust? 😵 I'll be speaking at @theleaddev London (June 1-2) about why "connecting AI to everything" doesn't work and what actually does when building tools that...
AI Is Redefining How We Build Software
The first recording from The Pragmatic Summit is available to watch for anyone: How AI is Reshaping the Craft of Building Software. With @thsottiaux (Head of Codex, OpenAI) and @vijayeraji (CTO of Applications, OpenAI) A very good one. Watch it here: https://t.co/apNHGykKbg...
AI Success Depends on Retrieval, Not Just LLMs
People think AI apps are just “prompt → answer.” Real AI products use RAG. DoorDash: retrieves past support cases + KB articles, adds guardrails, and an LLM judge checks quality. LinkedIn: uses a knowledge graph from tickets to solve issues faster. Bell: employees query...
Bespoke Software Era Highlighted at Prag
Fragments: security with OpenClaw, impressions from Pragmatic Summit, era of highly bespoke software, life-size pocket map https://t.co/IB6tYQu820
From IC to Product Leader: Identity Shift, Not Promotion
🚀 IC to Product Leader isn’t a promotion. It’s an identity shift. Happening TODAY. We’re hosting a high-signal fireside chat with Petra Wille — author of STRONG Product People — to unpack what really changes when you step into product leadership. 🗓 23 February ⏰...
Software Engineering Lives Beyond Code: Architecture Drives Business
They were right about software engineering being dead. Just not in the way people think. Software isn’t code. It’s architecture, tradeoffs, and understanding the customer well enough to know what not to build, and connecting technical decisions to real business outcomes. That isn't dead...
Product‑led Promises Often Mask Missing Accountability
For most companies, going "product-led" is a complete disaster, and they soon revert to a feature factory. It's a predictable pattern: • Product teams cheer. • Leadership steps back. • Accountability quietly disappears. • Then… it all blows up. WHAT COMPANIES THINK THEY ARE...
Closing the Loop: Feeding Production Feedback to Developers
ah, this is THE question: how to bring the feedback on changes back TO developers? i wrote this piece very recently on developer feedback loops in production: https://t.co/4GqKa1SCBA which includes this 3 min honeycomb demo from september: https://t.co/GNPVLI37YE
AI‑written Code Will Cause Outages without Proper Safeguards
This headline today reads shocking, but will soon enough be a nothingburger: When AI generates most code and more config files as well, most outages caused will be *technically* done by AI. In reality they happen thanks to inadequate verification/monitoring/rollback...
Trace Each Hop: Faster Debugging in AWS EKS
🌐 From Route 53 to Pod : The Real Network Flow in AWS + EKS Understanding Kubernetes networking isn’t optional in production. This diagram shows the complete request journey: Route 53 → IGW → ALB → Security Groups → Private Subnets → Ingress...
AI Writes Code Faster than We Can Review
▶ Reading unfamiliar code is exhausting. Now imagine that code is coming from an LLM that writes faster than you can think and doesn't take lunch breaks ◀ This 👆 perfectly captures the AI productivity paradox: code is written faster than...
Anthropic Alienates Open‑source Devs, Pushes Enterprise‑only Lock‑in
Anthropic has alienated devs who care about using open source tools, portability between vendors and using LLM subscriptions in the tool of *their* choice This all seems deliberate: Anthropic gunning at enterprise, not caring much if these devs go elsewhere/use other...
Expose All Product Surfaces to Empower Agentic Users
How well are you surfacing your product experiences to agents? Enjoying this Bengaluru Builder Day talk from @Swiggy's CTO that's looking at making all their surfaces available to agentic users. https://t.co/o6lDaMS5wd
AI's Easy; Scaling Software 100x Faster Is the Real Hurdle
What people seem to be struggling with the most is not the truly novel parts of building with AI, it's the table stakes. It's not the AI part that's so hard (turns out systems have never been wholly deterministic 😉), it's...
AI Productivity Tricks Don’t Scale to Real Team Workflows
Most AI productivity hacks you see on LinkedIn don't work. They work for one person, on their own laptop, with no one else involved. That's not your reality. Your reality is shared Google Docs with 14 contributors. Confluence pages that...
Martin Fowler’s Timely Advice & Guest Lineup Revealed
new post up: in which @martinfowler gives me the best advice ever at the worst possible time, plus... a rundown of the special guests to expect in the second edition of "Observability Engineering"! https://t.co/ZcaP1rzYib
Built Rails Infrastructure for Per‑Customer SQLite, Even When Plans Shifted
The original plan for Fizzy was to ship the SaaS version with one sqlite database per customer. That didn't end up happening, but we built all the Rails infrastructure needed to make it possible. Mike goes through it all here:...
Speed, Not Scale, Drives AI Agent Advantage
OpenClaw proved that one person can out-ship billion-dollar labs on agents. What does that say about where the competitive advantage actually is?
Physical AI: The Next Leap for Human‑Device Workforce
In this follow-on clip, I unpack why physical AI (and world models) is the next major leap. It introduces an entirely new kind of workforce: intelligent devices and humanoids working alongside people. The move from pilot to production won’t be...
Tech Stack and Agile Leadership Drive Banking Performance
A strong tech stack supported by change-ready leadership is essential for improved performance in today’s banking environment. Access the free report: https://t.co/0wah7wPuXN https://t.co/afm9zdd3tL
Bad Tools Drive Talent Loss and Competitor Advantage
Give engineers subpar tools, what could possibly go wrong? a) competitors out-ship AWS (eg Vercel, where Claude Code is used by so many and they are on 🔥) b) frustrated engineers interview elsewhere and take offers at places where Claude Code or...
Founders &
I was at a founders + CTOs event for 10-500 person startups. Founders and CTOs at these companies burn some of the most tokens and increasingly ship a bunch of code to production (when they did ~zero 6 months ago)
Basecamp Upgrades to Fully Compliant OAuth 2.0 Implementation
Basecamp has long supported OAuth, but our implementation was based on the (now ancient) pre-release spec, and it required hoops for modern clients. We've updated it to be fully compliant with OAuth 2.0 now. https://t.co/ixQWa4GmTH