Blog

Ai

37 posts tagged ai — the same founder-tested patterns, filtered.

Multi-Agent LLM Pipelines in Rails

Multi-Agent LLM Pipelines in Rails

Our production multi-agent LLM pipeline in Rails: eight RubyLLM agents on a 21-line base class, a 16-line loop, and the deploy that starved the database pool.
Evaluating LLM Agents in Rails

Evaluating LLM Agents in Rails

How a Rails team knows whether its LLM agents still work: bounded schemas at the boundary, and a per-iteration audit trail you can read a month later.
Debugging RubyLLM Agents in Rails

Debugging RubyLLM Agents in Rails

What a green Rails suite actually asserts when the thing under test is a hosted model: VCR matches on method and URI only, and the model can be retired.
Vibe Coding Is Disposable. Stop Shipping It.

Vibe Coding Is Disposable. Stop Shipping It.

Vibe-coded code is unmaintainable by design - patches don't stick, bugs return, the rebuild verdict lands. The generate-validate-kill workflow that prevents it.
AI Code Has an Owner Problem

AI Code Has an Owner Problem

The Linux kernel just made humans legally responsible for every AI-generated line. Your dev shop contract should too. Five clauses, one weekend audit, answer by Monday.
How We Replicated a Full Product Team With Claude Code Agents

How We Replicated a Full Product Team With Claude Code Agents

Solo Claude Code agents ship like solo developers - missing PM, designer, QA, customer voice. Here is the team-mode setup running Shape Up plus XP plus reflection-driven Kaizen.
The Quality Tax: AI MVPs Cost More to Fix

The Quality Tax: AI MVPs Cost More to Fix

Agencies pitch AI-built MVPs as faster and cheaper. The 2026 data says fixing them costs 2-3x what you saved. Here is the real founder math, with sources.
9 Seconds: An AI Agent Wiped a Live SaaS

9 Seconds: An AI Agent Wiped a Live SaaS

On April 25, 2026, an AI coding agent at PocketOS deleted the production database and its on-volume backups in 9 seconds. Here's what your dev shop should learn.
47 Startups Failed the Same Way

47 Startups Failed the Same Way

An Inc.com audit found 91% of failed startups had no tests, 68% had security holes. The 25-month decay timeline that predicts your codebase's fate.
Vibe Coding Crisis: Why AI Code Breaks

Vibe Coding Crisis: Why AI Code Breaks

Vibe-coded code looks fine until you try to change it. Five signals that tell you whether your codebase will survive its first year of paying customers.

Reading this because something is going wrong?

A free code audit gives you a written assessment of your codebase in plain English.

Get a Free Code Audit

Rated 4.8/5 on Clutch · you keep the write-up either way