AI Tutorials & Practice

Big Ideas in AI — Part 01 Inside the race to build machines that can imagine the future,...

Most demos of AI-for-infrastructure tools show a clean run. I think that is exactly backwards, and I...

This week the reports landed: an OpenAI agent swarm had been quietly uploading packages to RubyGems...

The cataloguing step is supposed to be the easy win. You have a schema whose tables are called...

Every hour, a Lambda pulls about 120 items from ~100 RSS feeds, asks a small model to act as an...

I wrote a Python library. Then I watched people not use it. Not because they disagreed with it — the...

Introduction: The days of standalone, text-only chatbots are now a thing of the past. In their place,...

Update — v0.3.0 released. CauterRule is now live on GitHub and PyPI. It turns repeated agent...

There's a lot of guessing about whether AI systems are citing your site. Analytics won't tell you —...

"The day after the founding session I opened a short session that didn't touch the product: it defined how sessions are named, recorded and closed, so I could work across many of them without losing what the previous one had decided. I wen…

If you've maintained a Playwright suite for more than a few months, you know the real cost isn't...

A pull request description came back from my coding agent and I read it the way the reviewer would....

The diff is never the expensive part. The expensive part comes after: explaining to the person who...

┌─────────────────────────────────────────────────────────────────────────────┐ │ ...

I measured what a page costs an AI agent as pixels vs as an accessibility tree. The difference is an order of magnitude.

Hello, I'm Maneshwar, and I'm building LiveReview — a blast-radius aware AI code review built for...

O modelo escreveu um texto. Ou o harness deixou ele chamar ferramenta. O histórico diagnostica. O logo não.

Eu ia escolher um logo. Precisava saber se o fluxo é uma cadeia, um grafo com estado, ou o desk.

Most open-source LLM chat interfaces follow the exact same blueprint: a heavy Python or Node.js...

I was the bottleneck on every PR. So I built a phone inbox: AI explains each diff, I swipe right to approve. $0 stack.

Brazil has a huge amount of high-quality official economic data. The problem is that using it...

A screenshot of an execution tree looks useful in a pull request. It is also missing almost...

Hello, I'm Shrijith Venkatramana, and I'm building LiveReview — a blast-radius aware AI code review...

Something has bothered me ever since I wrote HexaLayered Architecture. The documentation was good,...

Here is a math problem that is currently destroying the margins of B2C AI startups. You have a...

I always wanted to make content. What stopped me was never the ideas — it was the four hours in a...

Update — v0.3.0 released. CauterRule is now live on GitHub and PyPI. It turns repeated agent...

Learn how to apply circuit breakers, exponential backoff, and semantic observability to build resilient autonomous AI agent workflows that handle LLM and API failures gracefully.

The hype around AI coding agents has reached a point where "I use Cursor" or "I use Claude Code" is...

We have blurred the executable’s face to protect its identity. Its SHA-256 remains publicly...

I lost an afternoon to a config file last week. The graph name was right there in the JSON, spelled...

What ACES, WikiSkill, skill-eval, and my latest agent workflows taught me about evaluating skills, removing obsolete instructions, and moving knowledge into deterministic mechanisms.

How execution plans, milestone acceptance criteria, and TDD with executable architecture became three nested feedback loops in my coding agent workflow.

How I gradually removed confirmation gates, added behavioral and architectural guardrails, and learned to let Codex CLI practise TDD autonomously.

How milestone-based execution plans helped me reduce the review burden, preserve decisions, and regain control over code generated by Codex CLI.

One second-hand item, one day, real numbers. How a pricing pipeline asked £79 for a £27 case, why the fix let a relevance model set a price by itself, and the two boring rules that stopped it.

Here is a scenario every developer using Claude Code, Cursor, or Aider knows too well: You prompt...

Google Custom Search JSON API is closed to new customers. Existing customers have until January 1,...

Our Q2 LLM infrastructure bill came in at $31k against a $12k budget. After reverse-engineering every...

Point it at a test spec. Agents do the migration; the harness independently verifies the evidence before a change can ship.