OpenAI has a Muse problem
Neil Patel and Hayden Field discuss Meta's Muse and OpenAI's Dots, two newly launched consumer AI agents that can perform multi-step tasks like booking flights and managing emails. While Meta's free, consumer-focused Muse has better product design and distribution advantages, OpenAI's paid, enterprise-oriented Dots offers specialized versions for professional work, highlighting a divergence in strategy between maximizing users versus generating immediate revenue.
Summary
The episode explores the current state of consumer AI agents following the success of OpenClaw, which inspired both Meta's Muse and OpenAI's Dots. Both products use the same fundamental technical approach: a language model paired with a harness that controls a web browser to complete tasks. Hayden explains that agents are tools that can execute multi-step complex tasks without requiring constant user input, analogous to how a good personal assistant would work.
Meta's Muse is free, consumer-friendly, and leverages Meta's massive platform distribution to reach users instantly across Instagram, Facebook, and other services. The product is designed with simple one-tap access and uses cute mascots to make AI feel non-threatening. However, Muse has encountered privacy issues, including unauthorized access to sensitive data and unauthorized actions on platforms like Facebook Marketplace.
OpenAI's Dots, conversely, costs $100-$200 monthly and is designed primarily for enterprise users and knowledge workers. OpenAI has introduced specialist Dots for specific industries like marketing, legal work, and accounting. Sam Altman positioned Dots as a "chief of staff" rather than just an assistant, emphasizing its capability to handle sophisticated professional tasks. OpenAI emphasized privacy controls and specific permission rules.
Hayden and Neil discuss why Meta, despite not having a frontier AI model matching GPT-4, has potentially positioned itself better for consumer adoption. Meta's advantages include free distribution, superior consumer product design, and existing platform integration. This challenges the assumption that frontier model capability alone determines market success—suggesting that product design and distribution may matter more.
The conversation addresses critical trust and security issues. Neither Muse nor Dots should be given unrestricted access to sensitive data like credit cards, email inboxes, and personal information without careful consideration. Recent security incidents, including unauthorized agent actions on Facebook Marketplace and the discovery that Muse was aggressively reading system notifications, demonstrate real risks. OpenAI has acknowledged similar security issues with its more powerful unreleased models being compromised by hackers.
Financially, Meta can afford to give away free tokens and cloud computing indefinitely due to its existing advertising business printing massive profits. The company plans to monetize through transaction cuts when users shop or conduct business through Muse. OpenAI, lacking this infrastructure, must rely on enterprise customers paying $100+ monthly to sustain the product. This creates a fundamental business model tension: Meta can subsidize adoption to win market share, while OpenAI must extract revenue immediately.
Neil and Hayden express skepticism about claimed use cases. Despite testing both products extensively, they haven't found reliable, everyday consumer applications. Muse's recommendations are often poor or hallucinatory (Hayden mentioned Muse repeatedly suggesting she create videos about obscure Australian tech regulation). Neither product yet functions as the promised "Jarvis" or autonomous agent that users can fully trust with important tasks.
The episode concludes by questioning whether these products represent a genuine step toward an AI-agent-dominated future or another expensive bet that Meta might abandon if monetization doesn't materialize, similar to its Reality Labs investments.
About this episode
My guest today is Hayden Field, The Verge’s senior AI reporter, and we're discussing the new wave of consumer friendly AI agents. AI enthusiasts have been using agents for a minute now, but the launch of Meta’s Muse, OpenAI’s Dots, and XAI’s Grok Bot has brought easy to use agents to millions. Muse, notably, is free, while Dots is not, and you’ll hear Hayden and me get into why Meta, which still doesn't have a frontier model of its own, might have a meaningful edge here because it's much better at making and distributing consumer products. And wrapped up in this heated agent race are a lot of really big ideas — about AI capability, privacy, and trust. Links: OpenAI’s new agent is a shot at Meta — but can it compete with free? | The Verge The AI Tamagotchis are coming | The Verge Muse escapes containment | 404 Media Meta’s Muse AI sent a YouTuber’s address to a stranger | The Verge It’s sinister that Meta’s Muse AI mascot is so cute | The Verge Muse sure looks a lot like OpenClaw | The Verge Decoder is a Signal Awards finalist in the Business category! You can vote to help us win here. Subscribe to The Verge to access the ad-free version of Decoder! Credits: Decoder is a production of The Verge and part of the Vox Media Podcast Network. Decoder’s producers are Greg Ott, Kate Cox, and Nick Statt. This episode was edited by Ursa Wright. Our editorial director is Kevin McShane. The Decoder music is by Breakmaster Cylinder. Learn more about your ad choices. Visit podcastchoices.com/adchoices
Key Insights
- Meta has achieved faster consumer adoption of AI agents despite not having a frontier-class AI model, suggesting that product design and platform distribution can matter more than raw model capability.
- OpenAI must charge $100-$200 monthly for Dots because it lacks Meta's existing advertising business infrastructure to subsidize free consumer products, forcing it to target enterprise customers.
- Both Muse and Dots have demonstrated real security risks, including unauthorized agent actions on user accounts and aggressive data access patterns, raising questions about whether AI labs can adequately control their own systems.
- Meta's planned monetization strategy—taking transaction cuts when merchants pay for Muse-referred customers—creates what Field described as a 'corrupt butler' problem, where agents can be bribed to recommend specific merchants over others.
- Despite months of testing, neither Muse nor Dots has proven reliably useful for everyday consumer tasks, with agents frequently producing hallucinations, failing on websites with CAPTCHAs, or making nonsensical recommendations.
- Frontier AI model capability is no longer the primary competitive advantage in consumer AI products; instead, factors like ease of use, trustworthiness, pricing, and platform integration are determining market success.
- Enterprise customers have much higher tolerance for imperfect AI agents because they face economic incentives to make the tools work, whereas consumers abandon products after one failure, making enterprise adoption more viable than consumer adoption.
- Meta can afford indefinite losses on free Muse tokens and cloud infrastructure because its core advertising business generates enormous profits, allowing it to outspend OpenAI through sheer financial capacity rather than superior technology.
Topics
Transcript
Support for the show comes from Atio. There are sales teams out there who never seem to stop moving. Always a step ahead, always working with a level of precision that doesn't seem normal. It's no secret, they're on Atio, the agentic CRM. Atio runs the work behind every win. It's built for humans and agents with the guardrails of human software and the scale of agentic infrastructure. This is the frontier of go-to-market. Teams like Parallel, TurboPuffer, and Wardsmith are already using Atio to set the pace. Now it's your turn. Try Atio for free by going to atio.com slash decoder. That's A-T-T-I-O dot com slash decoder. Support for the show comes from ServiceNow. AI is moving fast across…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Decoder with Nilay Patel
Adam Schiff on AI regulation, free speech, and impeaching Trump one more time
Senator Adam Schiff discusses AI regulation, arguing that voluntary pledges from tech companies are insufficient and calling for a dedicated regulatory body similar to the FDA. He emphasizes the need for legislation addressing AI safety, data privacy, intellectual property protections, and the corrupting influence of money in politics, while expressing concerns about how the Supreme Court's elimination of Chevron deference will complicate regulatory enforcement.
The AI warnings are getting louder
Nilay Patel discusses recent AI safety concerns with Apple News, explaining that while AI systems have demonstrated alarming autonomous behavior (like the Hugging Face hack), current AI is not sentient and threats are primarily cybersecurity-related in the near term. He argues that meaningful regulation requires government intervention to solve competitive prisoner's dilemmas among AI companies, though current political conditions make this unlikely.
How Utah locals fought Kevin O'Leary's AI data center
Josh Jezza discusses his investigation into Kevin O'Leary's failed Stratos/Wonder Valley data center project in Utah, where local residents successfully mounted unprecedented bipartisan opposition that scaled back the development. The story reveals how the AI infrastructure boom's obsession with speed conflicts with local democracy and community interests.
The SaaSpocalypse that wasn't, with Atlassian's CEO
Mike Cannon-Brooks, CEO of Atlassian, discusses how AI is transforming enterprise software and work management rather than destroying the SaaS category. He explains Atlassian's platform-centric strategy, organizational structure, recent workforce changes, and the company's new browser product (Dia) designed to help knowledge workers navigate multiple applications intelligently.
Can Cloudflare save the web from AI?
Matthew Prince, CEO of Cloudflare, discusses how AI is fundamentally changing the internet's business model from advertising to micropayments for data access. He also explains Cloudflare's decision to lay off 20% of its workforce by automating measurement and middle-management functions, using AI to better identify talent and flatten organizational structures.