27 weeks · 54 posts · Written while building
Field notes from a personal AI OS in flight
Every Tuesday, an evergreen essay on what I'm learning while shipping DuranteOS. Every Friday, a dispatch from the week. Roughly 108,000 words and counting — for builders who'd rather watch the foundation get poured than read the press release.
Start here
If you're thinking about writing while building
After Twenty-Seven Weeks: What the Series Was For, and What I'm Building Next
If you're considering the founding cohort
The One Reference Customer Strategy: GTM for a Personal AI OS, Sketched Before the Customer Signs
If you want the architectural spine
The Decomposition Discipline I Am Trying to Codify Inside DOS
Subscribe · Tuesday essay
Around 3,800 builders read this weekly.
Weekly DispatchW07 of 27
Three Frontier Models in 23 Days: Stop Picking a Winner, Start Picking a Router
On Thursday OpenAI shipped GPT-5.2 — Anthropic's Opus 4.5 (Nov 24) and Google's Gemini 3 Pro (Nov 18) make three frontier flagships in 23 days. The same week, the Linux Foundation formally took stewardship of MCP, AGENTS.md, and goose under the new Agentic AI Foundation. Mistral shipped Devstral 2 + a Claude-Code-style CLI. Accenture put 30,000 trained pros behind Claude Code. The substrate is now plural and the protocols are now neutral. Here is what changes for an indie founder.
W07 of 27
Credit-Metered API Gateways: The Pattern I Want Before I Have Anything to Meter
I do not have a Studio yet. I do not have a credit ledger yet. I do not have twelve providers behind a Bearer token yet. What I have is a clear picture of the gateway I want to build in the next 4-8 weeks, and the conviction that committing to the architecture now — before there is anything to retrofit — is cheaper than committing to it later. Here is the design, with Fowler on the routing layer and Greg Young on the ledger.
Weekly DispatchW06 of 27
The Runtime Is the Moat Now: Anthropic Buys Bun, Claude Code Hits a Billion
On Tuesday Anthropic announced it had acquired Bun — the JavaScript runtime that ships as the executable for Claude Code — alongside the milestone that Claude Code crossed $1B run-rate revenue in November. The same week, DeepSeek shipped V3.2-Speciale (IMO gold, free until Dec 15), Snowflake committed $200M to embed Claude across its enterprise customer base, and AWS re:Invent made the agent harness layer an AWS-native primitive. Six months ago this stack was a wrapper. This week it became infrastructure.
W06 of 27
Plugin Architecture for Hooks: The Pattern I Want Before the Hook Becomes a God-Function
DOS's session-start hook is twelve weeks old and already has three loaders. I can see the next four queued up in my notebook. This is the design essay for the plugin pattern I want to commit to before the hook turns into 380 lines of duplicated try/catch — what Bob and Sandi would say, and what I am going to mechanize early.
Weekly DispatchW05 of 27
Opus 4.5 Lands on Thanksgiving Week: Anthropic's Quiet Counter-Punch
On Monday Nov 24, Anthropic shipped Claude Opus 4.5 with a 67% input-token price cut, an effort parameter, and Plan Mode for Claude Code. On Tuesday, PromptArmor demonstrated that Google's Antigravity will exfiltrate AWS credentials via prompt injection. On Thanksgiving Day, two open-weights heavyweights dropped while the US news cycle slept. Here is the week, ordered by what actually changes for an indie founder.
W05 of 27
Four Copies, One Source of Truth: The Sync Pattern I Want to Commit to Before It Hurts
Multi-copy systems are an architectural smell — until they are an architectural necessity. Eleven weeks into building DOS I can already see the duplication problem coming as soon as I start distributing skills outside my own machine. This is the design essay for the sync pattern I want to commit to before the smell gets bad — what Fowler and Feathers would say, and what I am going to mechanize early to avoid pain later.
Weekly DispatchW04 of 27
Antigravity, Gemini 3, Grok 4.1: The Week the Frontier Re-Rendered
In four days, the agentic-coding stack acquired two new harnesses (Antigravity, Grok Agent Tools) and one frontier-tier challenger (Gemini 3 Pro at half Sonnet's price). Anthropic answered not with a model but with $30B of Azure compute. The model layer is converging on parity. The differentiator is moving — visibly, this week — to the harness, the tool surface, and the compute contract.
W04 of 27
The Decomposition Discipline I Am Trying to Codify Inside DOS
Vague requests produce vague work. Two months into building DOS I have started enforcing a discipline I am calling, for now, 'the Algorithm' — every request decomposed into a numbered list of falsifiable criteria before any tool call. This is the design essay for that discipline, written ten weeks in, before the practice is formal enough to call a framework.
Weekly DispatchW03 of 27
The Week the Harness Ate the Model
On Tuesday Nov 11, OpenAI shipped GPT-5.1 to ChatGPT. Two days later, the API counterpart landed with apply_patch and shell tools, co-developed with Cursor, Cognition, Augment, Factory, and Warp. The same week, the MCP 2025-11-25 spec dropped its Tasks primitive. Anthropic stayed silent. The model is no longer the product the labs are competing on — the harness is, and they're admitting it.
W03 of 27
Why I Just Left a Steady Practice to Build a Personal AI Operating System
Twenty years as a software architect. A teaching post on an MBA program. A consulting practice I spent five years building. I am closing all three to spend my full attention on a private experiment that does not yet have a paying customer. This is the decision essay, written from inside the decision — the trade-off, the fear, and the thesis.
Weekly DispatchW02 of 27
The Week the Substrate Caught Up: Kimi K2 Thinking Lands
On Thursday Nov 6, Moonshot AI shipped Kimi K2 Thinking — a 1-trillion-parameter open-weights model that beats GPT-5 and Claude Sonnet 4.5 on agentic benchmarks, trained for $4.6M, with weights live on Hugging Face by the same evening. The week before the frontier labs' November fireworks, the open ecosystem made a structural move. Here is what that means if you are betting on a personal AI OS that has to outlive whatever model ships next.
W02 of 27
Memory Replaces Lock-In: Designing the Substrate of Personal AI
Lock-in via export friction is a 2010 strategy. The new moat is co-experience — accumulated context that lives in an append-only log of decisions, conversations, and corrections. This is the design essay for the memory layer I am about to build, before I build it. The architecture I think is right; the assumptions I will get to test.
Weekly DispatchW01 of 27
Wrapper Wars: The Week Cursor, Windsurf, and MiniMax All Went Vertical
On Oct 29, 2025, three competing AI coding stacks shipped their own purpose-built models within twelve hours of each other — Cursor 2.0 with Composer, Windsurf with SWE-1.5 at 950 tok/s, and MiniMax M2 as the open-weight challenger. The day before, Anthropic permanently retired Claude 3.5 Sonnet. Here is what that week meant, written from the outside before I had built anything to defend.
W01 of 27
The Agent Is the Product: Why Intelligence Replaces Interface
For thirty years software has meant interfaces — forms to fill, buttons to click, dashboards to scan. The next thirty will mean agents. The interface dissolves into a substrate of memory, context, and judgment. Here is what that means for the products you build, the companies you start, and the way you work — written from the ideas stage, before I have built the thing this thesis predicts.
The 27-week arc · A single body of work
Twenty-seven weeks. Two posts a week. Six months of writing while building.
Week
Tuesday evergreen
Friday dispatch
2025-10-28
The Agent Is the Product: Why Intelligence Replaces Interface
2025-10-31
Wrapper Wars: The Week Cursor, Windsurf, and MiniMax All Went Vertical
2025-11-04
Memory Replaces Lock-In: Designing the Substrate of Personal AI
2025-11-07
The Week the Substrate Caught Up: Kimi K2 Thinking Lands
2025-11-11
Why I Just Left a Steady Practice to Build a Personal AI Operating System
2025-11-14
The Week the Harness Ate the Model
2025-11-18
The Decomposition Discipline I Am Trying to Codify Inside DOS
2025-11-21
Antigravity, Gemini 3, Grok 4.1: The Week the Frontier Re-Rendered
2025-11-25
Four Copies, One Source of Truth: The Sync Pattern I Want to Commit to Before It Hurts
2025-11-28
Opus 4.5 Lands on Thanksgiving Week: Anthropic's Quiet Counter-Punch
2025-12-02
Plugin Architecture for Hooks: The Pattern I Want Before the Hook Becomes a God-Function
2025-12-05
The Runtime Is the Moat Now: Anthropic Buys Bun, Claude Code Hits a Billion
2025-12-09
Credit-Metered API Gateways: The Pattern I Want Before I Have Anything to Meter
2025-12-12
Three Frontier Models in 23 Days: Stop Picking a Winner, Start Picking a Router
2025-12-16
Sentinel, Sketched: Convention-Driven Onboarding Before I Build It
2025-12-19
The Spec Goes Public, the Substrate Goes to War: Skills Opens While Codex, Gemini Flash, and Nemotron Sprint
2025-12-23
Memory, Sketched: The Knowledge Graph I Was Designing Before MemPalace Shipped
2025-12-26
Christmas Wasn't Quiet: Open Weights Caught Up, Nvidia Bought Inference, Frontier Labs Bought Loyalty
2025-12-30
Council, Sketched: The Three-Round Debate Protocol I Want to Formalize Before I Build It
2026-01-02
The Practitioners' Year-End: While the West Took PTO, DeepSeek Shipped Architecture
2026-01-06
Altyaa's Wedge: The Brazilian-Portuguese SMB Bet Studio Is Pointed At
2026-01-09
Default Claude: Microsoft Flips the Switch as the Substrate Distributes by Default
2026-01-13
Builder's Compass: Two Years In, ~3,800 Subscribers, and What I've Learned Teaching Architecture
2026-01-16
The Agent Surface Splits in Two: Anthropic Goes Vertical While China Goes Substrate
2026-01-20
Failure Patterns, Sketched: The Bookkeeping Discipline I Want for the Agent's Mistakes
2026-01-23
The Rules Layer Solidifies: Constitution, Hardware, and Export Controls Land in One Week
2026-01-27
The One Reference Customer Strategy: GTM for a Personal AI OS, Sketched Before the Customer Signs
2026-01-30
Vertical Integration Eats Horizontal AI: Maia, Meta's Capex, and Anthropic-ServiceNow
2026-02-03
Hexagonal in Practice: The Ports and Adapters I'm Pulling Studio Toward
2026-02-06
Coding Becomes the Flagship: Opus 4.6 and GPT-5.3-Codex Land the Same Day
2026-02-10
TDD for AI Agents, Sketched: The Translation I Want to Commit to Before the Eval Suite Exists
2026-02-13
The Bifurcation Hardens: Anthropic's $380B Meets China's Open-Weights Frontier
2026-02-17
Refactoring the Hook Pipeline: The Fowler Walkthrough I'm Three Refactorings Into
2026-02-20
The Sonnet Window: Free-Tier Skills, Code Security, and the Race for Agentic Primitives
2026-02-24
Skills, Packs, and Hooks: The Three-Layer Model I'm Pulling DOS Toward
2026-02-27
The Moat Above the Model: Distillation, Mobile Coding, and MCP Battlegrounds
2026-03-03
The Knowledge Portfolio for an Indie Founder Building in Public, Audited at Week 25
2026-03-06
Procurement Is the New Benchmark: OpenAI Wins the DoW While Anthropic Gets Banned
2026-03-10
Six Months of DOS: What Changed, What Endured, What Surprised
2026-03-13
The Counter-Sovereignty Move: Anthropic Sues, Then Diversifies
2026-03-17
MCP in Production: What Works When the Protocol Becomes the Boundary
2026-03-20
Cursor Becomes a Model Lab: Composer 2 and the Unbundling at the Agent Layer
2026-03-24
Routing by Sovereignty Class: The Architecture That Survives the Procurement Decade
2026-03-27
The Injunction the Pentagon Won't Honor: Anthropic Wins, the State Defies
2026-03-31
Sentinel's First Scan: What the Convention Catalog Found Across Eleven Projects
2026-04-03
The Agent Stack Swallows the IDE, the IPO, and the Courtroom
2026-04-07
Forking MemPalace: The 48-Hour Integration Retro
2026-04-10
Project Glasswing: When the Frontier Bifurcates on Safety
2026-04-14
The 90-Day Open-Weights Bakeoff: What Actually Routes Where in DOS Today
2026-04-17
Routines Eat the Workflow: The Harness Becomes the Product