ai
What's Happening With AI Token Prices, and What It Means for Your Budget
A practical look at why AI token prices can feel both expensive and rapidly cheaper, and how falling inference costs change what customers should expect from AI products.
ai
Does your team use different AI agents? Here’s how to standardize your products.
If your team uses a mix of Cursor, Claude Code, and Copilot, things can get messy fast. In this article, we explain how to set up a system to standardize your products, keep your design system consistent, and ensure every AI agent plays by the exact same rules.
ai
Understanding AI Code Fast: A 60-Second Habit for Institutional Memory
Why teams using AI-generated code still need lightweight verification, and how a 60-second micro-review habit with git-lrc can build institutional memory without reading every line.
ai
A New Method for Stable Software: Micro Code Reviews for the AI Era
Why AI-generated code needs lightweight verification, and how micro code reviews with git-lrc can improve software stability and developer comprehension.
ai
The Subsidy Era Is Over: A Reality Check on AI-Powered Dev Tool Pricing
A reality check on AI-powered dev tool pricing, shrinking subsidies, workload-based costs, and predictable pricing for code review tooling.
ai
Why Agentic Engineering Must Replace Vibe Coding
A critique of vibe coding and a case for agentic engineering built on verification, feedback loops, accountability, and engineering rigor for serious software systems.
design
Responsible Design: Shaping Large-Scale Consequences by Organizing Agents
A reflection on AI accountability, individual responsibility, and why careful design is the only viable way to shape outcomes in complex agentic systems.
ai
Searching Billions in Seconds: How HNSW Solved the Scale Problem
Discover how HNSW solved the scale problem for Approximate Nearest Neighbor Search by creating a multi-layered graph system that can search through billions of items in milliseconds.
ai
The future belongs to those who can refute AI, not just generate with AI
In an era of machine-generated conjectures, engineering advantage will belong to those who practice Popper's falsification at scale.
ollama
Diagnose & Fix Painfully Slow Ollama: 4 Essential Debugging Techniques + Fixes
Frustrated with laggy Ollama? Try out these debugging techniques
ai
Building a Bayesian Spam Classifier from First Principles
We work through the original Bayesian Spam Classifier paper from Microsoft Research. And then make it practical by applying to the Enron spam dataset
ai
From Logic to Linear Algebra: How AI is Rewiring the Computer
For decades, computers were built for logic -- branching decisions, compilers, databases. But AI flipped the script: today's machines are being rebuilt for one thing above all else -- multiplying matrices at massive scale