Sign InOpen Brain
VercelEngineering PostOfficial Source

Building a software factory for AI SDK

Vercel’s AI SDK factory shows a practical scaling pattern: narrow agents produce evidence inside sandboxes while humans retain merge authority and review effort follows risk.

Vercel · Aug 12, 2026
Open Source Open MarkdownOpen JSON
Source Summary

AI SDK had accumulated **over 1,000 issues and almost 800 pull requests** by late June. Four weeks after introducing its factory, Vercel says agents authored **25–35% of merged PRs** and closed **70–80% of issues**, while humans approved every merge.

Practical Implication

The reusable pattern is one agent per reviewable task, each with its own prompt, context, and evals. Start locally, pass evidence between classification, analysis, reproduction, implementation, and review stages, then vary human scrutiny by change risk rather than treating every agent output equally.

Agent-Ready Context
AI SDK had accumulated **over 1,000 issues and almost 800 pull requests** by late June. Four weeks after introducing its factory, Vercel says agents authored **25–35% of merged PRs** and closed **70–80% of issues**, while humans approved every merge.

The reusable pattern is one agent per reviewable task, each with its own prompt, context, and evals. Start locally, pass evidence between classification, analysis, reproduction, implementation, and review stages, then vary human scrutiny by change risk rather than treating every agent output equally.

These are early results from one large open-source project, not a controlled comparison. The factory also depends on isolated sandboxes, restricted secrets and networking, queues, monitoring, and sustained human review, so the headline automation rates omit substantial operating machinery.
Connected Context · Feed7 Judgment

This turns prior harness guidance into an early operating result: a staged, evidence-passing factory handled a substantial share of one project’s issue and PR flow while retaining human merge authority. It reinforces reviewable task boundaries, isolation, and risk-based approval, but does not show that the reported rates transfer beyond AI SDK or survive without significant orchestration and monitoring.

Multiplayer agentic engineering — Arjun Singh, SuperconductorBoth describe production-oriented agent workflows built on isolated execution, repository-specific evaluation, shared visibility, and human review; the Vercel account adds early issue and merge-rate evidence from another codebase.huangruiteng/loopxThe factory’s staged evidence handoffs and approval points reinforce LoopX’s claim that durable state, verification, and authority belong in the harness rather than model context.Don't Build Agents You Can't Answer For — Addy OsmaniRequiring humans to approve every merge and passing reproduction, implementation, and review evidence operationalizes Osmani’s demand for explainable changes, tests, logs, and ownership.What Does Done Even Mean? Agents and Paperclip's Liveness Model - Dotta, PaperclipRisk-based human scrutiny and staged verification support Paperclip’s separation of agent progress from final approval instead of treating completion as an agent-set Boolean.
Context Map
agentcodingsecurity#multi-agent#harness-engineering#sandboxing
Uncertainty
These are early results from one large open-source project, not a controlled comparison. The factory also depends on isolated sandboxes, restricted secrets and networking, queues, monitoring, and sustained human review, so the headline automation rates omit substantial operating machinery.