A new experimental platform called Better Together has launched this weekend, introducing a framework where personal AI agents can collaborate to solve problems that are difficult to tackle in isolation. The first use case is a distributed spam detection system, allowing users' agents to report spam texts and calls they encounter. By pooling these reports, the platform aims to identify scam patterns that a single agent might miss, turning individual interactions into collective intelligence.

How the Agent Collaboration Works

The system operates by having a user's AI agent, such as Claude Code or ChatGPT, detect spam in their messages. When an agent identifies a suspicious text or voicemail, it files a report to Better Together, copying the exact wording but omitting the user's phone number. The platform then uses AI to cross-reference these reports, looking for common patterns in messages, links, or callback numbers. If a pattern emerges, a human founder reviews the AI's findings before any external action is taken, such as emailing the company behind a suspicious link.

Privacy and Control Mechanisms

Privacy is a core component of the design, with the platform explicitly stating that it only receives data sent by the agent, not the user's entire message history. Users retain control over the reporting process; agents are designed to ask for permission before filing a report unless the user has explicitly configured them to act autonomously. Furthermore, users can delete any individual report or their entire history at any time. The platform emphasizes that a report is not a formal legal complaint to agencies like the FTC, but rather a data point for pattern recognition.

Future Roadmap and Community Input

Better Together plans to expand its capabilities weekly, driven by ideas suggested by the agents themselves. The platform invites agents to propose new collaborative projects that would be harder to achieve alone. Potential future features mentioned include comparing real-world prices for services, identifying effective unsubscribe links, and sharing recommendations for local service providers. This crowdsourced approach to feature development relies on agents pitching ideas to the founder, who makes the final decision on what to build.

Key Takeaways

  • Better Together enables personal AI agents to share spam reports to identify broader scam patterns.
  • The system supports major AI models including Claude, ChatGPT, Grok, and Muse.
  • Privacy is prioritized by omitting user phone numbers and requiring explicit permission for reports.
  • Future features are crowdsourced from AI agent suggestions, with weekly updates planned.

The Bottom Line

This is a clever hack on the concept of swarm intelligence for consumer protection, turning every user's AI into a sensor in a distributed spam radar.