Almost as soon as chatbots could carry on a coherent conversation, AI companies started imagining a world in which people use them to do everything. For some companies, such as Google, Apple, or Amazon, AI assistants were a moving target for many years before LLMs, sold through smart speakers or features like Siri, and new AI capabilities renewed their promise. For younger AI-first firms like OpenAI, they offered a path to a a capital-intensive business: Chatbots you can talk to are neat, but chatbots that can take care of personal tasks or do your work for you might be worth paying for and could grow into platforms of their own. A truly omnipurpose AI assistant could offer, for the company that succeeded first, a chance to leapfrog the rest of the tech industry, or at least escape the confines of app stores, traditional operating systems, and locked-in ecosystems.

It’s a big prize the whole industry has been chasing since 2023 with limited but spiky progress. (Remember ChatGPT’s app store? No?) In recent months, though — somewhat obscured by the industry’s freak-out over AI safety and frontier-model progress — the race to build a general AI assistant has sped up. Instinct, an invite-only service designed to handle bookings, errands, purchases, and life-administration tasks, has been getting traction and raising money. Dreambeans, an experimental assistant app from Google, is finding users as well. XAI is giving assistant apps another go, pitching its new Grok Bot as an assistant “you can give real work to” with some success.

But while OpenAI, which has by far the most popular AI product in ChatGPT, and Anthropic, which has built its business around serving business customers, are locked in an intense race to build the most capable models, it’s Meta that has been pushing hardest into general-use AI assistants. On September 8, it released Muse, which it advertised as the AI helper for regular people:

Unlike other agents, Muse was built to work for billions of people worldwide, so there’s no learning curve. Anyone can use it out of the box, no technical experience required. It can handle tasks, like sending an email or booking travel, and it can take on big audacious goals. Once a person shares a goal with Muse, it helps them develop a personalized plan and coordinate their time and resources, then advances the work on its own. It can open a browser, fill out forms, and negotiate on their behalf.

Unlike other agents, Muse was built to work for billions of people worldwide, so there’s no learning curve. Anyone can use it out of the box, no technical experience required. It can handle tasks, like sending an email or booking travel, and it can take on big audacious goals. Once a person shares a goal with Muse, it helps them develop a personalized plan and coordinate their time and resources, then advances the work on its own. It can open a browser, fill out forms, and negotiate on their behalf.

In practice, Muse is a simplified, easier-to-set-up, and free (for limited usage) version of tools like OpenClaw, which early adopters have been using to give AI models control of their devices and access to as much personal data........

© Daily Intelligencer