Notes

Notes from building SpielOS

I write about agent systems, product failures, context, workflows, evaluations, and the decisions behind SpielOS.

Mostly the parts that broke before they worked.

27 notes 106 topics

Sep 6, 2026

How to Automate Accounting and Bookkeeping With AI

How to automate accounting and bookkeeping with AI: what you can automate, how the loop works, benefits with real numbers, costs and ROI, and a step-by-step rollout.

#accounting automation Read more
Aug 30, 2026

How to Automate Recruitment With AI: CV Screening and Candidate Matching

How to automate recruitment with AI: candidate screening, CV screening, and candidate matching on the tools your team already uses, with a human still making the calls.

#recruitment automation Read more
Aug 23, 2026

My AI Company Generated 2,000 Qualified Leads While I Slept

No prospecting team. Zero dollars spent on lead gen tools. See how an AI worker researched, verified, scored, and rejected leads all weekend — and how you can install the same worker yourself, free.

#AI lead generation Read more
Aug 4, 2026

Every Workflow I Reuse Is a Skill. I Made 10 Public.

10 production-grade skills I actually use daily: video creation from HTML, content writing, Persian translation, SEO, analytics, design systems, and more. Full repo, full scripts, full templates.

#Open Source Read more
Aug 1, 2026

The Waitlist That Almost Didn't Work

You use AI to build fast. But the field that decides whether your launch has a market is usually the least verified thing you own. The story of wiring a waitlist to an isolated Supabase table with RLS, so the signups actually arrive.

#Ops and Architecture Read more
Jul 31, 2026

SEO Is a Byproduct of Architecture

You already use AI to generate marketing pages. The problem is rarely the copy — it is the structure. Refusing to invent a marketing taxonomy and letting the product architecture define the sitemap was the fastest SEO win on this site.

#Ops and Architecture Read more
Jul 31, 2026

The Dictionary That Fixed Translation

If your AI output is inconsistent, the reflex is to blame the model. Most of the time it is not the model. It is the missing glossary — the contract between your product and the language you are generating. The story of turning a failed translation pass into a living dictionary.

#Ops and Architecture Read more
Jul 30, 2026

RTL at the Design-System Level

If you serve Persian speakers — or any second language — i18n is a system problem, not a string problem. Fonts, direction, spacing, icons, and tokens all change. The story of making RTL the default, not the exception.

#Ops and Architecture Read more
Jul 29, 2026

DeepSeek V4 Official Launch: Everything We Know About the Real V4 in July

DeepSeek V4 official stable launch confirmed for mid-July 2026. Peak-valley pricing, DSpark 85% speed boost, legacy API retirement July 24. Full breakdown.

#deepseek-v4 Read more
Jul 27, 2026

DeepSeek V4: The Model That Rewrote the AI Price War — Full Breakdown

DeepSeek V4 rewrites AI economics: $0.14/M input tokens, 1M context default, MIT-licensed open weights, 3206 Codeforces rating. Full benchmark comparison vs GPT-5.5 and Claude Opus 4.7.

#AI Market Read more
Jul 27, 2026

The Migration Audit

If you generate code with AI, you already know the shape of this story: the thing that compiles is not the thing that works. Migrating from Jekyll to Astro was easy. Knowing what the migration broke was the real work — 45 pages, one green build, two real audits.

#Agents, Loops & Evals Read more
Jul 1, 2026

The AI War Is Not About Free Models. It Is About Collapsing the Price of Intelligence

The real AI war is not OpenAI vs DeepSeek or closed vs open source. It is the U.S. premium cloud meter against China's attempt to collapse the price of intelligence.

#AI Market Read more
Jun 30, 2026

Tokens Per Sheep: The AI Metric Nobody Asked For

OpenCode Go users in New Zealand used 40.9B tokens last week. That is 1,734 tokens per sheep. We did the math nobody asked for.

#AI Market Read more
Jun 21, 2026

Is GLM-5.2 Cheaper Than the Frontier?

A working breakdown of LLM API pricing across Anthropic, OpenAI, Google, and Z.ai, at every reasoning effort. Where the popular claim breaks down, where it could technically be true, and what the data says about intelligence.

#AI Market Read more
Jun 19, 2026

Why Every Technical Founder Needs a Content Pipeline, Not a Content Strategy

Content strategy is the wrong abstraction for builders. You don't need a plan. You need a system that turns building into publishing.

#Founder Distribution Read more
Jun 18, 2026

3 Things Shipped to Spiel OS This Week

Spiel OS is a pipeline that turns a build session into a publishable post. The gap that kept showing up was not the writing. It was the publishing, the last 20% that takes 80% of the energy.

#Founder Distribution Read more
Jun 14, 2026

Positioning Is Not a Copy Change. It Is a System State

You updated your bio. Your system didn't. Here is why your old positioning is still running in files you forgot existed, and how to fix it in an afternoon.

#Founder Distribution Read more
Jun 10, 2026

How I Built a Developer Content Pipeline in 72 Hours

A developer content pipeline workflow built in 72 hours. 4 blog posts, 3 LinkedIn posts, 2 state machines, ~85 queued drafts. One system that makes output compound.

#Founder Distribution Read more
Jun 8, 2026

From Sessions to Posts: The 4-Layer Stack That Makes It Work

How to turn raw AI conversations into publication-ready content. A technical breakdown of the session-to-post pipeline with eval gates and content architecture.

#Ops and Architecture Read more
Jun 8, 2026

Your Session Logs Are Your Content Engine

How session logs become your most valuable content asset. A technical deep dive into capturing, processing, and publishing from AI conversations.

#Ops and Architecture Read more
Jun 8, 2026

Gates Not Models

Why your AI agents produce garbage and how to fix it with validation gates. A practical guide to building reliable content pipelines with evaluation-driven development.

#Agents, Loops & Evals Read more
Jun 8, 2026

Spiel OS is Now Open Source: An AI Writing Pipeline for Builders

An open-source AI writing pipeline that turns build sessions into publishable content. 4 files, 2 loops, 19 commands, MIT licensed. Clone it, run it, own it.

#Agents, Loops & Evals Read more
Jun 8, 2026

Content Pipeline: The Full Stack

A complete technical breakdown of the content pipeline that turns raw thinking into published posts. From session logs to live content in 16 minutes.

#Ops and Architecture Read more
Jun 8, 2026

Every Decision Is Content

Why the decisions you make building your product are your best content source. How to capture and publish decision logs as blog posts.

#Ops and Architecture Read more
Jun 8, 2026

From Declarative Rules To Agentic Loops

Why your LLM agents ignore half your rules. Learn the agentic AI workflow architecture that replaces fragile prompts with state-machine loops that actually work.

#Agents, Loops & Evals Read more
Jun 7, 2026

How I Rebuilt My Blog in 4 Hours With a Jekyll Automation Pipeline

A Jekyll blog automation pipeline that turns vault drafts into live posts with one command. From empty demo to brand blog with 2 posts shipped the same night.

#Founder Distribution Read more
Jun 6, 2026

How I Automated My Content Pipeline With an AI Second Brain

Build an AI content pipeline for developers that turns build sessions into publishable drafts. Open-source, 4 files, runs from your terminal.

#Founder Distribution Read more
Browse by topic (106)