Gotanda Style: Do AI Agents Really Need Meetings?

Short summary

Gotanda Style is a stigmergy-based coordination pattern where AI agents avoid direct communication, instead leaving structured signals (pheromones) in a shared environment to reduce token overhead. This approach scales to large codebases by converting coordination costs to state updates, with real-world validation on a 200K-line Python repository automating Sentry alerts to GitHub pull requests. The pattern enables conflict detection and nuanced decision-making beyond simple signal aggregation.

• •Direct agent communication creates token overhead; structured signals in a shared environment (stigmergy) offer a lighter-weight alternative
• •Real-world validation: 200K-line Python repo with Sentry alerts → pheromone field → GitHub issues → PRs, reducing coordination latency
• •Signal system tracks strength, decay, and conflicts, enabling nuanced decision-making (contested areas vs. strongly positive/negative zones)

Generated with AI, which can make mistakes.