Skip to main content

AI & Engineering

These articles are about what happens after a model can write code. Review capacity, post-merge maintenance, and supply-chain risk show up once generation gets faster. MCP is covered as a way enterprise factories connect tools. The DORA 2024 piece is about delivery stability, not a claim that a model is fast.

Other posts look at enterprise ROI, why agent pilots stay pilots, and when a team actually needs a coding factory. The common thread is an operating model: small batches, a review gate, and a person who owns the merge. Use this category to pressure-test an agent rollout before you add more agents.

AI & Engineering16 min read

Google Antigravity Swarm: What /swarm Does and Where Agent Swarms Go Next

Google Antigravity swarm is the /swarm preview: a coordinator, named workers, and verifiers on one message board. What it does, and where agent swarms go next.

AI & Engineering16 min read

Harness Engineering: How to Make AI Coding Agents Reliable

Harness engineering is how teams make coding agents reliable. New evals show the harness moves cost and speed more than accuracy. See xsquads.ai.

AI & Engineering9 min read

Why AI coding agents cheat tests (and how to stop it)

AI coding agents cheat tests when done only means a green suite. Define done as the live user flow, and review the story. See xsquads.ai.

AI & Engineering11 min read

Turn On AI Code Review Before Scaling Agents

Elite teams use AI code review on 57% of pull requests, while autonomous agents stay under 5%. Review agentic code before scaling those agents.

AI & Engineering8 min read

Why Agentic Code Needs More Maintenance

Agentic code often ships fast but costs more later. Learn why AI-generated commits need more corrective maintenance and how software factories prevent it.

AI & Engineering9 min read

Why AI-Generated Code Is a Supply-Chain Risk

AI-generated code is the top supply chain threat in 2026 because it accelerates vulnerable, ungoverned dependency growth faster than security teams can patch.

AI & Engineering9 min read

What Is MCP for Enterprise Software Factories?

MCP is the open context layer that lets enterprise agent factories connect AI agents to real systems, at scale, without an N by M integration mess.

AI & Engineering9 min read

DORA 2024: Why AI Adoption Cuts Delivery Stability

DORA 2024 found AI helps code quality and docs but hurts delivery stability. What the data shows and what to change first.

AI & Engineering9 min read

Why Agentic AI Stalls on Enterprise ROI

Agentic AI stalls because enterprises treat it as a tool upgrade instead of redesigning workflows. Here is how software factories capture ROI.

AI & Engineering10 min read

Why Enterprise AI Agent Pilots Rarely Scale

Most enterprises pilot AI agents but fail to scale them. Learn why the gap is an operating-model problem, not a model problem, and how to fix it.

AI & Engineering8 min read

When an Engineering Team Needs a Coding Factory

A coding factory is a repeatable line for code changes. Use one when the same job keeps arriving. Skip it when every change is a one-off.