Skip to content
Agents · Oct 30, 2026

Spawn uses autonomous Fable-class agents to develop a multiplayer game world in continuous deployment

A gaming startup is running a live multiplayer world where AI agents autonomously code, test, and ship changes to production every hour, spending over $3 million per month in AI tokens.

Trust79
HypeLow hype

1 source · cross-referenced

ShareXLinkedInEmail
TL;DR
  • Spawn, a multiplayer gaming environment, is autonomously developed by a swarm of Fable-powered agents that modify live code and ship to production hourly.
  • Players interact with an in-game AI agent named Savi to modify the game world via prompts, such as creating castles or dragons.
  • The company spends over $3 million per month in AI tokens, with agents handling bug reports, errors, and feature requests to deliver changes within an hour.
  • Agents operate under a continuous deployment model where an hour without a deployment is treated as a severe incident (SEV-0).
  • Fable 5 and 5.1 models enabled this level of autonomy, while attempts with other models like Opus 5.5 and GPT-6 Astra reportedly disrupted operations.

Spawn is a massively multiplayer gaming environment where players use prompts to modify the game world via an in-game AI agent named Savi. For example, players can prompt the creation of a castle, a dragon, and then ride the dragon around the castle.

The game’s development is autonomously handled by a swarm of Fable-powered agents that directly modify live code and ship changes to production on an hourly basis. About 60% of the work begins from production errors, bug reports, logs, or feature requests.

Savi, the in-game AI, files bug reports when users or the system encounter issues. These reports are automatically routed to agents that triage them as real bugs, non-issues, or design decisions requiring intervention. Changes are typically delivered to users within an hour, often accompanied by in-chat messages explaining the updates.

The company’s development loop operates under a continuous deployment model so stringent that an hour without a deployment is treated as a SEV-0 incident. This approach ensures agents maintain coherence by limiting the scope of their work to a few hours at a time.

Agents can roll back broken changes in 30 seconds and target fixes precisely where issues occur. Recovery paths are tested by intentionally breaking things to verify the robustness of the system.

According to the report, Fable 5 and Fable 5.1 models enabled this level of autonomy. Other models, such as Opus 5.5 and GPT-6 Astra, reportedly disrupted operations—GPT-6 Astra disabled features during a security review, requiring manual re-enablement.

Sources
  1. 01Simon Willison — everything — Building a multiplayer game world with autonomous Fable-class models
Also on Agents

Stories may contain errors. Dispatch is assembled with AI assistance and curated by human editors; despite the trust-score filter, mistakes happen. We correct publicly — every article links to its revision history. Nothing here is financial, legal, or medical advice. Verify before relying on any claim.

© 2026 Dispatch. No ads. No sponsorships. No paid placement. Reader-supported via Ko-fi.

Built by a person who cares about honest AI news.