Skip to content
K = 0.731

Simon Willison Examines Spawn’s Autonomous Agent Development Workflow

1 reports 1 reporting sources 2 hours agoNew

Understand the event

AI overview

Simon Willison examined Jacob Sansbury’s account of building the multiplayer game Spawn with autonomous Fable-powered agents, describing a workflow that Sansbury says deploys to production hourly and supports targeted rollbacks in 30 seconds. The agents reportedly consume more than $3 million in tokens per month. Sansbury described automated bug triage and said the team treats an hour without a deployment as a SEV-0, its most severe incident category. Willison tried Spawn on his phone, asking its in-game agent, Savi, to create a castle and a dragon he could ride. He sees games as a promising setting for exploring coding agents, while questioning whether this development approach is suitable for higher-stakes software.

Generated by AI from reports · Updated 1 hours ago

Latest development10/30 01:00 (UTC+8)
Building a Multiplayer Game World With Autonomous Fable-Class ModelsSimon Willison

Read next

More changes in today’s briefingGoogle DeepMind announces Gemini 4 Carbon with cybersecurity and long-horizon capabilities

Report timeline

Follow the reports to explore different sides of the event.

10/30
  1. Simon Willison
    Building a Multiplayer Game World With Autonomous Fable-Class Models

    Simon Willison examines Jacob Sansbury’s account of developing Spawn with autonomous Fable-powered agents reportedly costing over $3 million in tokens per month. Willison tried the multiplayer game on his phone, prompting its in-game agent, Savi, to create a castle and a dragon he could ride. Sansbury describes hourly production deployments, automated bug triage, and targeted rollbacks in 30 seconds; Willison sees games as a promising setting for exploring coding agents while questioning this approach’s suitability for higher-stakes software.

Keep up with AI, one email a day

Model releases, price changes and compute developments, with original sources. Free, confirmed by email, and easy to unsubscribe. Read a past issue

Event heat trend

There is not enough continuous observation data to plot a trend yet.