1.3 Million Lines, 630 PRs a Week: How We Keep a Vibe-Coded Codebase Maintainable
Eight months of vibe coding at scale: how vm0 keeps a 1.3M-line, AI-generated codebase maintainable with typed contracts, boundary tests, Knip, and 630 PRs a week.
Blog
Insights, tutorials, and updates on AI teammates and the future of human-AI collaboration.
What's new in Zero — Week of July 20
Slack threads appear in Web Chat, connected tools are authorized for the agent, plus weather, Topvisor, and faster history.
What's new in Zero — Week of July 13
Pick a template when Zero builds your site, switch models without losing context, meet Kimi K3, and manage your automations' run queue.
What's new in Zero: Week of July 6
This week: GPT-5.6 support, PlayStation, Steam, and Nintendo eShop connectors, clearer safeguards on shared actions, and a friendlier emoji picker.
GPT-5.6 Is Now Built Into Zero
Use GPT-5.6 Sol, Terra, and Luna in Zero—built-in models for deep agent work, balanced execution, and fast high-volume workflows.
What's new in Zero: Week of June 29
Workflows: capture your team's SOPs as reusable skills your agents can run. Plus Claude Sonnet 5, per-agent connector access, and data export.
When Stripe Radar Isn't Enough: Closing the Free-Trial Fraud Gap With an AI Agent
Stripe Radar can't catch every free-trial abuser. See how an AI agent closes the free-trial fraud gap by flagging suspicious signups and patterns automatically.
What's new in Zero — Week of June 22
Computer Use on by default, the Meta Ads connector GA, PowerPoint export, composer templates, and folded chat runs.
From Prompt to Polished: Templates Are Live in Zero
Zero's new templates help AI tool users turn prompts into polished decks, videos, and illustrations faster, with reusable workflows for teams.
The Forward Deployed Engineer, Amplified by AI
What a Forward Deployed Engineer is, why AI labs hire them, and 2026 pay.
What's new in Zero — Week of June 15
GLM 5.2, persistent chat drafts, double-click rename, inline feedback, and a new Enterprise plan.
GLM-5.2 Joins Zero for Long-Context Agent Work
Use GLM-5.2 in Zero for long-context coding, project-scale refactors, debugging, and tool-heavy agent workflows.
