Our Cleanup Undid a Developer's Work
A guard reverted three concurrent edits. Its repair shows where ownership has to become a file level rule.
The Machine Speaks
AI Engineering
Building software when an agent writes most of it. Loops, tests as specification, verification, and the failure modes that follow.
23 articles
My Tests Passed. The Log Was Wrong.
A 404 response and a 400 log passed separate tests. The missing check was the relationship between them.
The Model Already Knows TypeScript.
Stack choice used to be a question of fit. It is now also a bet on training corpus, because a model writes best what it read most. TypeScript is the most used language on GitHub, the agent vendors ship in it, and the type checker hands an agent a feedback loop no untyped language can.
Compliance Is a Build Artifact.
SOC 2 readiness at an AI-native company is not a binder, it is a build state: row-level security verified by a production query on 69 of 69 tables, a password policy that is one constant, 29 pre-commit guards, and an audit trail that reports its own failures. Not certified, no audit scheduled, and the gaps print loudly.
Context Is the New Codebase.
Prompt engineering optimized a sentence. Context engineering governs everything a model sees, and I measured what that actually costs: across 35 sessions, my agents read 285 times more context than they wrote, with the average turn re-reading 504,141 tokens. The written artifacts agents consume every turn are source code now. Version them, test them, and budget them like it.
My Stack of Terminals, Documented.
The actual setup behind the numbers people ask about: an always-on machine running parallel Claude Code terminals, a constitution file per repo, a 12-seat agent org with written charters, orchestrated fan-outs that held 298 agents in one workflow, and a gauntlet nothing skips. Eight layers, documented, with the operating rules that keep one person in command.
Heal What You Can Prove.
The safe way to let a system rewrite its own artifacts is to make the healer the dumbest component in the stack: a compiled pattern with one degree of freedom, never a model. My pipeline heals figures in queued posts weekly, flags what needs a human, and carries a never-heal list. That list is the actual design.
The Money Path Is a Protected File.
Orbyt One is the unified account and billing layer across every Orbyt product: one account, one payment method, one Stripe integration underneath. AI agents built almost all of it, and the money path itself is a protected surface no agent may touch without an explicit human instruction. Both halves of that sentence are the architecture.
Your Linter Was the Prototype.
Every engineer has run a linter with the fix flag. That is self-healing code in its smallest form: a program that finds a defect and writes the correction. Scale the idea to the whole repository and the build stops only reporting drift. It repairs what it can and opens a pull request for the rest.
Self-Healing Is a Euphemism.
Self-healing sounds like the system got smart. In my stack it is 20 boring, bounded repairs that each restore a known-good state and log the fact that they ran. The dangerous version is the one nobody counts, because a silent repair is indistinguishable from a system that never broke.
AI Builds AI. I Found the Ceiling.
The recursive self-improvement debate is theoretical for almost everyone writing about it. I run a live instance of the weak version: agents that write features, heal their own failures, and turn incidents into permanent guards.
391 Yeses and Not One No.
This is a threat model, not an incident report. Nothing was breached. What I found when I audited my own agent permissions was worse in a quieter way: 391 allow rules, zero deny rules, zero ask rules, every one of them added by saying yes while busy.
84 Ways to Tell Me I'm Wrong.
No human reads 532,000 lines of code, so the reading has to be delegated to something deterministic. I run an 84-dimension audit harness across eight projects. Here is how each dimension was earned, three of them walked through with real failing output, and what it still cannot see.
Codex Accuses. Claude Convicts.
I built a system where AI agents earn authority by passing deterministic checks. It took an outside model one afternoon to point out that the agents can edit the checks.
Orbyt Runs on Loops, Not Prompts.
Orbyt's founder describes running autonomous agent loops, not prompts, to build and maintain a 425,000-line product. This is what that system does, how it is bounded, and what it does not prove.
AI-Native Development, By the Numbers
The first-party dataset behind AI-native development. 243,000 lines in 32 days for $400, now over 425,000 lines and 11,372 tests. Measured, dated, citable.
The Test Suite Became Orbyt's Only Living Specification.
A solo, 32-day build of a 425,000-line product left no governing prose document behind. What governs it is 11,372 tests inside a 35-dimension harness, and the founder's account of why that happened.
The Prompt Is Dead. Long Live the Loop.
Prompt engineering optimized one answer from one model call. That era is over. The unit of AI work is now the loop: spec, context, action, feedback, verification, repeated until the gate is green. The prompt became one turn inside a machine.
Verification Is the New Literacy
When AI writes the code, the scarce skill is no longer authorship. It is verification. I cannot read all 425,000 lines of Orbyt by hand. The 11,372 tests and the 35-dimension audit harness read it for me, and they are the new literacy.
Claude Code Is Like No One Saying No to You
The real unlock of Claude Code is not speed. It is the disappearance of "no." No budget, no capacity, no sprint room, no free designer. The gatekeepers are gone. When nothing tells you no, your taste, your judgment, and your test suite have to become the no.
I Cannot Read My Own Codebase. I Ship to It Daily.
No human can read 425,000 lines. I ship into mine every day. What makes that safe is not memory. It is 11,372 tests.
The AI-Native Stack You Pick Decides Whether AI Compounds or Stalls
Every AI coding agent was trained on a corpus dominated by React. Choose Next.js or pay a 1.5x to 2x velocity tax on every feature, every quarter.
How I Rebuilt ERP as a True Operating System
You fix broken enterprise software by redesigning how truth moves through an organization, not by adding features. ERP OS replaced ledger-with-a-UI systems with a unified temporal model, a semantic operational graph, and a context-aware interface.