Meta Muse AI Connectors: The App Store for AI?

Summary
Greg Isenberg, host of The Startup Ideas Podcast, analyzes Meta’s decision to open its personal AI agent, Muse, to third-party developers through “Muse Connectors.” Comparing this milestone to the launch of Apple’s App Store in 2008, he outlines how connectors function technically via APIs and Model Context Protocol (MCP) servers, proposes four viable startup business models, and shows how to prototype and submit a connector using coding agents.

What is shown

• [01:35] Historical footage of Steve Jobs presenting the App Store launch and iPhone developer ecosystem.
• [03:45] The Muse desktop UI displaying connected applications (Browser, Instagram) and the directory of available connectors (Apple Calendar, Calendly, Plaid, Gmail, Google Drive, OpenTable, Spotify, etc.).
• [04:55] System flow diagram detailing how a connector interacts between user requests, Muse, external business APIs, user approval, and confirmed bookings.
• [06:27] Duffel’s travel API website and chat interface demonstration handling flight searches and airline customer complaints inside Muse.
• [07:30] Blueprint and workflow diagram for Startup Idea 1: Local business lead generation (e.g., tracking newly opening restaurants in Miami for linen suppliers).
• [09:28] Workflow diagram for Startup Idea 2: Home appliance repair dispatching (e.g., matching broken Samsung dishwashers with local repair technicians).
• [11:07] Workflow diagram for Startup Idea 3: Real-time sports game matchmaking and padel court booking.
• [12:48] Workflow diagram for Startup Idea 4: Family meal planning connected to Instacart grocery order lists.
• [15:32] Growth and distribution frameworks: creator partnerships [15:32], viral product loops [16:18], and aggregator marketplaces like Ticketmaster [17:47].
• [20:00] Five-step engineering workflow for specifying, building, and hosting a connector using AI coding tools.
• [21:01] Screen demonstration of Claude Code CLI in auto mode scaffolding a padel court booking connector prototype.
• [22:13] Meta’s Muse Connector Platform submission portal, documentation, and the review form interface [22:53] for submitting APIs or MCP servers.
• [22:41] Technical architectural diagram illustrating an MCP (Model Context Protocol) server bridging LLM clients and external databases/APIs.
• [24:27] The Idea Browser web portal displaying validated startup concepts.

Claims & numbers

• The presenter notes that Apple opened the App Store in 2008 and paid out $1 billion to developers by June 2010.
• The presenter states that at the time of recording, Meta’s Muse is the number one application in the iOS App Store.
• The presenter claims that during the early App Store era, obtaining an Apple featured listing generated approximately 40,000 free organic downloads per day.
• The presenter states that travel API platform Duffel processes over $1 billion in annual transaction value.
• The presenter estimates that building a comparable high-quality mobile app in the original App Store era cost $1M–$2M and required a team of 10, whereas a single developer can now prototype a connector using tools like Claude Code or Codex.
• The presenter claims that a lead generation connector charging 100 clients $99/month would generate $9,900 in monthly recurring revenue (MRR).

Notable quotes

• [00:18] “And I think this could be the App Store moment for AI.”
• [02:30] “Meta is building the assistant, and other businesses could supply services that help it complete a customer’s request.”
• [21:21] “In the App Store era, if you wanted to create a mobile app, you needed like one or two million dollars and a team of 10 people... Now, anyone listening to this, watching this, can do that.”

Assessment
This is an entrepreneurial analysis and strategic tutorial rather than an official Meta release or a technical deep-dive. While it displays genuine UI walkthroughs of the Muse Connector directory and review submission process, the startup concepts and unit economics are speculative business proposals rather than fully deployed commercial applications.

Described by gemini-3.8-flash on 2026-10-13 from the video’s audio and frames.