OpenAI's Dots vs. Meta's Muse in under 17 minutes

The video compares two new personal AI agent applications: Muse by Meta and Dots by OpenAI. Both function as always-on cloud-based agents that users can name, customize, and interact with across multiple platforms. The reviewer notes that Dots appears to be a rushed competitive response to Muse's rapid adoption at the top of the Apple App Store, even crashing during its announcement at OpenAI's Dev Day.

Regarding fundamental capabilities, both apps use different underlying models (Dots uses GPT-4o Astra while Muse uses Meta's Spark 1.3), provide cloud computing resources, offer integration ecosystems with single sign-on connectors, can execute scripts and custom APIs, and create documents, presentations, and interactive dashboards. Dots integrates directly with ChatGPT and Claude/Codex if users have those apps and grant access, while Muse has its own independent tool ecosystem.

The key differentiator is user experience. Muse provides a more polished interface with clearer features: editable memory and soul files (persona settings), scheduled task management, an ideas tab for exploring use cases, and the ability to organize multiple conversation threads by topic. The interface also includes an approvals tab, schedule view, and a personalized feed. Dots, by contrast, feels oversimplified and rushed—the reviewer struggled to find basic functions like scheduled task management and found the interface confusing and unintuitive.

Communication options differ: Dots integrates with Slack (allowing team members to message the agent in three clicks) and supports voice calls with low latency. Muse offers WhatsApp integration and allows viewing agent interactions there, though replies must come through the main app. Muse can create podcasts and videos from content, while Dots emphasizes direct voice communication.

Pricing is another distinction. Muse offers a free tier with up to 100 million tokens per week with no current user payment required, and optional paid upgrades. Dots requires a paid ChatGPT subscription to create an agent, though temporary promotional pricing doesn't count agent usage against subscription limits.

The reviewer criticizes Dots for multiple issues: buggy functionality, poor integration with Codex projects (agents create tasks but don't properly post them to project management tools), inconsistent documentation where the agent gives incorrect instructions about how to use it, and an overall sense that the product wasn't ready for launch. The reviewer tweeted that "Dots should have been called Bugs."

Regarding model capability, GPT-4o Astra is more intelligent overall, excelling at coding and complex software automation, while Spark 1.3 handles routine intellectual work, document creation, and general assistance adequately and slightly faster. For general personal assistant work, Spark suffices; for technical automation, Astra is superior.

The reviewer predicts convergence: these three agents (Muse, Dots, and Grokbot) will eventually adopt similar feature sets and interfaces, with differentiation based on underlying models, ecosystem integrations, and pricing. The primary decision factors for users will be whether they're already invested in ChatGPT/Codex, Meta platforms, or other ecosystems, and which communication channels (Slack, WhatsApp) matter most to them.