Artificial Intelligence

LLM Tracing vs Evals: Which One Tells You Why Your Agent Broke

Your agent gave a bad answer and you cannot say why. Tracing tells you what happened on one request. Evals tell you whether quality moved across many. They answer different questions, and if you only build one this quarter, build the one that makes the other actionable.

19 Aug 2026

Containers vs Cloudflare Computer: There Is Not Enough Compute for One Box Per Agent

Nearly every agent framework decided that one agent means one container. Cloudflare says there are not enough computers on earth for that to work, and has flipped it around: the agent lives in a tiny sandbox with a filesystem that stays behind, and a container is only borrowed when Linux is truly needed. Here is the argument in plain words, why Cloudflare made it, and where I would use it.

12 Aug 2026

Graph Engineering: Every Edge You Draw Takes a Decision Away From the Model

Agent graphs are state machines with a new name on them. The engineering question is not what a node is, it is which routing decisions you take away from the model and which you leave to it. The rule I apply: only draw edges for decisions you can make better in advance than the model can at runtime.

9 Aug 2026

Semantic Caching for AI Requests, Explained Simply

Your users ask the same question in different words, and you pay for every one. A semantic cache answers from meaning instead of exact text. Here is how it works, in plain language, with a working version you can build in an afternoon.

5 Aug 2026

The Most Honest Piece of Documentation Hugging Face Ships Is Its Logo

The company that hosts most of the open weights on the internet chose a smiley face with hands as its logo. That looks like a joke about its origins. I think it is the clearest statement of the product strategy anyone in AI infrastructure has made, and it explains why the enterprise content now sitting on that site feels slightly off.

19 Jul 2026

Why I Route Every AI Call Through OpenRouter Instead of Direct Endpoints

I migrated gazar.dev off direct Gemini and OpenAI endpoints onto OpenRouter as a single gateway. One key, one request shape, one bill, and switching models is now a string change. Here is what that bought me and what it cost.

17 Jul 2026

What I Learned From DoorDash's AI Assistant Architecture

DoorDash's eval said Gemini Flash was worse than Sonnet. It wasn't. Four things I took from their engineering series on grounding, agent memory, and evals.

14 Jul 2026

Soon Everyone at Your Company Will Be an Engineer

AI is tearing down the wall around engineering. Why everyone becomes an engineer, org charts flatten, and every function turns into an engineering role.

26 Jun 2026

Ship and Learn Are Two Different Metrics

AI defaults close your task, they do not keep you sharp. Used without intent it bills you in cognitive debt. The posture that keeps you learning while you still ship.

17 Jun 2026

The Broken Window Theory of Code, and Why Cheap AI Rewrites Don't Save You

One unfixed broken window lowers the standard for everything that follows. AI is making rewrites cheaper than ever, which raises an uncomfortable question: are we still going to understand our code, or just keep rebuilding it?

3 Jun 2026

Agents Who Code and Generative UI: Notes from AI Native Engineers London

At the second AI Native Engineers London meetup, the talks were about agents that ship real PRs, generative UI for MCP apps, and design systems in the loop. Here is my honest report from the night.

29 May 2026

Supervisor Agent Architecture: What Makes It Work

I have been reading deeply about supervisor agents: how they connect to RAG, how memory actually works, human-in-the-loop, dynamic agent generation, and what it costs to run them properly.

16 May 2026

MCP: Context Is Everything — Notes from a Building with MCP Event

14 May 2026

Engineer Role Shift: From Implementer to Orchestrator

The engineer role is shifting from writing code to orchestrating AI agents. Learn what architecture, coordination, and quality evaluation look like now.

5 May 2026

Traditional SDLC vs Agentic SDLC: What Actually Changes

Agentic SDLC vs the traditional software development lifecycle: what actually compresses, what stays hard, and what it means for senior engineers.

5 May 2026

Multi-Agent Topologies: Choosing Between Supervisor, Pipeline, and Broadcast

Most engineers building multi-agent systems reach for "just add another agent" before they've thought about what topology they're building. That's usually...

1 May 2026

From Prompts to Persistence: Agent Memory, Context, and Memory Engineering

Most agent demos are amnesiac by design. You send a message, the agent responds, and the next time you open a session it has no idea who you are. That is ...

30 Apr 2026

Running Gemma 4 Locally for Your Next Feature: A Practical Decision Guide

I built a commit message generator locally as a playground first, instead of reaching for a hosted API, so the model stays on my Mac (M4 these days), it cos...

11 Apr 2026

The Day 2 Problem: Why Vibe Coding Is Creating a Maintenance Crisis and What Good Engineers Do Instead

I was reviewing a PR last week where a checkout flow had grown to 847 lines in a single component. The author had used an AI assistant to build out severa...

10 Apr 2026

17 Claude Code Token Management Hacks (Tier 1 to Tier 3)

In the past week or so, people have been complaining about hitting their Claude Code limit insanely fast. Claims like one prompt that used to cost 1% of t...

6 Apr 2026

Anatomy of the .claude/ folder: The Control Center for AI-Enhanced Projects

When working with AI coding assistants, organization is everything. A clean, structured approach to managing instructions, rules, and specialized agents c...

29 Mar 2026

OpenClaw: The Self-Hosted AI Assistant That Treats Intelligence as Infrastructure

Most AI assistants live behind someone else's API. You send a message, it hits a server you don't control, and a response comes back through a pipeline yo...

8 Mar 2026

Token Economics: How to Cut LLM Cost Without Making Your Product Worse

Words you need

2 Feb 2026

Chunking Is Product Design (And Most RAG Systems Prove It)

Words you need

2 Feb 2026

Agentic Systems in Production: Agent Interfaces, Orchestration Patterns, and Observability

A chatbot answers questions. An agent takes actions.

15 Dec 2025

Embracing AI-Augmented Development: How AI is Changing the Way We Code

Six months ago I tried writing a service without any AI tooling. No Copilot. No Cursor. No ChatGPT in a side tab.

23 Mar 2025

Linear Regression in Artificial Intelligence

Linear regression is the "hello world" of machine learning. It predicts a number based on input data by finding the best straight line through your data p...

26 Mar 2024

A Guide to Creating AI Assistants with OpenAI: Unleashing Intelligent Task-Specific Models

I wanted to analyze my bank transactions from CommBank and ING. Not with a spreadsheet. With an AI that could answer questions about my spending patterns ...

4 Jan 2024

OpenAI APIs: From ChatGPT to Image Generatio

I wanted to understand what OpenAI's APIs could actually do. Not the marketing -- the real capabilities, the costs, and the gotchas.

30 Dec 2023

Machine Learning Use Cases in Future

Machine learning is already everywhere. Most people just don't notice it. Every recommendation, every autocomplete, every fraud alert -- there's a model b...

15 Oct 2023

Machine Learning Amazon SageMaker

Machine learning has a three-step lifecycle: prepare data, train a model, deploy it. Most tutorials focus on the first two. The third is where teams get s...

14 Oct 2023

Object Tracking with OpenCV

Object tracking means locating the same object across successive frames of a video. Detection finds it once. Tracking follows it over time.

14 Oct 2023