BLOG
Product updates, engineering deep dives and notes from the field — written by people who actually ship the code.
Subscribe and get one new post about code, design and shipping — straight to your inbox. No filler, no growth hacks.
You're almost in. Check your inbox to confirm.
No spam. Unsubscribe anytime.
BLOG
Product updates, engineering deep dives and notes from the field — written by people who actually ship the code.
Subscribe and get one new post about code, design and shipping — straight to your inbox. No filler, no growth hacks.
You're almost in. Check your inbox to confirm.
No spam. Unsubscribe anytime.
June 25, 2026
Adding paths to .gitignore and deleting them in a new commit only removes files from the latest snapshot of your branch. Old commits on GitHub still store every blob you ever pushed. To actually erase files from a public repository you must rewrite history with git filter-repo or an orphan branch,…
June 24, 2026
A non-technical solo founder can ship an MVP in 2026 with vibe coding : GitHub → Cursor → Next.js (or mobile tooling) → Supabase → Vercel, then commit and push for auto-deploy. If you have never been near software development , you will still have to pick up new basics: repos, deploy URLs, env…
June 20, 2026
Most MVPs do not need two founders on day one. They need five jobs covered: problem insight, product judgment, build, distribution, and enough runway discipline to stop gold-plating . Solo can work when one person owns the critical path and buys the rest with contractors and tools. The co-founder…
June 19, 2026
We publish letscodeit.dev blog posts from a separate GitHub repository. The Next.js app fetches markdown at runtime with a one-hour cache, syncs slug metadata into Postgres on every build, and writes some assets back through the GitHub Contents API . There is no admin UI for long-form articles. Git…
June 17, 2026
GEO (generative engine optimization) is how you structure and publish content so AI chatbots and answer engines quote your pages when someone asks a question. Classic SEO chases blue links in Google. GEO chases citations inside ChatGPT, Perplexity, and other answer products. We spent several weeks…
May 28, 2026
When you send a message to an LLM, you are not paying for characters or words. You pay for tokens . A token is the unit the model splits text into before processing: every major model API bills on tokens. For an individual user, the language of a request changes what you spend-or how fast you burn…
May 20, 2026
An LLM is a text predictor: you send words in, it predicts the next token until the reply is done. An agent wraps that same model with tools and a plan so it can search, write files, or call APIs. A subagent is a helper agent spun up by another agent for one slice of a bigger job. Same stack,…
May 13, 2026
Anthropic Academy is Anthropic's free learning platform on Skilljar . As of June 2026, it hosts self-paced courses on Claude chat, the API, Claude Code, MCP, plus role-specific AI fluency tracks. You do not need a paid Claude subscription to enroll. Sign up, pick a course, and work through modules…
May 6, 2026
This glossary defines 30+ AI and LLM terms in plain English: tokens, context windows, RAG, embeddings, fine-tuning, plus agent vocabulary. Each entry has a short definition plus an analogy, the way I'd explain it to a colleague who builds software but skipped the ML lecture series. Skip terms you…
April 29, 2026
AI will not replace developers in the foreseeable future. It automates boilerplate and tests, plus first-pass debugging, but someone still has to define the problem and own trade-offs. That same person takes the pager when production breaks. After two years of headlines about the end of…