Skip to content
Looking for someone to build your mobile app?See my mobile work
Writing

Blog

Notes on building for the web and mobile: performance, architecture, AI, and the craft of shipping.

2
series
15
articles

Series

· Multi-part builds, in order
Series8 partsComplete

Agent loop

Building a coding agent from scratch, one loop at a time, every part in runnable code.

Jun 2026 — Aug 2026≈ 3 h total read
Read the series →
  1. 01What an agent loop is (and isn't): state, action, stop29 Jun · 12 min
  2. 02The simplest loop that works: a write-test-fix agent, step by step30 Jun · 24 min
  3. 03The sandbox in an agent loop: from the model's text to a reliable observation30 Jun · 21 min
  4. 04Who writes the agent's tests: three ways to define what's correct30 Jun · 24 min
Series10 partsIn progress

Deep Agents

Building a website auditor with deepagents, one harness layer per post: tools, filesystem, permissions, sandbox, context, skills, and delegation.

Latest · Part 017 Aug 2026
Start with part 01 →
  1. 01What an agent harness is (and when not to use one): deepagents vs LangGraph7 Aug · 14 min
  2. 02Your first deep agent: create_deep_agent with frontier and open-weight modelsComing soon
  3. 03Custom tools and MCP in deepagents: Python functions, LangChain tools, and MCP serversComing soon
  4. 04Threads, checkpointers, and human-in-the-loop in deepagents: approving tool calls before they runComing soon

More writing

· Standalone posts
Illustration of Spec-Driven Development: a specification turns into acceptance criteria and from there into the earthquake list of a mobile app
AI5 Sep 202623 min read

Spec-Driven Development in React Native: an MVP that lists the world's earthquakes, specified before writing code

I applied Spec-Driven Development to a React Native MVP that lists the world's earthquakes: the specification decides what goes in the list, and its acceptance criteria become the tests.

Illustration of a voice agent: a call comes in, a backend resolves dates and options, and a confirmed appointment comes out on the calendar
AI4 Sep 202625 min read

Voice agent that books appointments: ElevenLabs, Cal.com and a backend that never lets the model guess

How I built a voice agent that books 30-minute appointments with ElevenLabs and Cal.com, and why the Fastify backend is what keeps the model from inventing times.

Data annotation illustration: a person trains a robot by approving a labeled example with a check mark while the training signal flows toward the robot
IA25 Jul 202613 min read

Data annotation jobs: what labeling and reviewing AI responses involves

What data annotation jobs and AI response review are: the tasks involved, how they feed model training (RLHF), and how annotator quality is measured.

Illustration of an agent fleet: a prompt fans out to several agents in parallel, each in its own worktree
AI6 Jul 202616 min read

Orca: the ADE for orchestrating a fleet of coding agents in parallel

I tried Orca, the open-source Agent Development Environment (ADE), to run a fleet of coding agents in parallel—Claude Code, Codex and more, each in its own isolated worktree.

Model routing illustration: a request is split across several models, with most of the traffic going to the main model
AI29 Jun 202625 min read

Model routing with OpenRouter and DeepSeek to cut costs without losing quality

How I routed models in production with OpenRouter, using DeepSeek for the bulk of traffic and premium models only where it matters, cutting LLM costs without degrading quality.

Illustration of a news pipeline: several RSS feeds converge into a flow that filters, rewrites with AI, and publishes a draft
Automation29 Jun 202616 min read

I automated a news site with n8n: from 20 RSS feeds to WordPress drafts

How I built an n8n pipeline that reads 20 RSS feeds, deduplicates with Postgres, curates and rewrites with tiered AI, and leaves the article as a WordPress draft with an image and a Telegram alert.