Top Tech Investor Explains The Best AI Assistants in 16 Minutes — Kevin Rose
In a Nutshell
Kevin Rose describes how new proactive AI assistants like Instinct, Muse, and Grokbot have shifted from reactive chatbots to autonomous agents that manage personal tasks, coordinate schedules, and handle commerce through deep integrations with email, calendars, and services. These systems create platform lock-in by storing comprehensive user context and preferences, while AI-mediated commerce threatens destination sites like Amazon as transactions move entirely into conversational interfaces. The key advantage lies with established companies over startups for privacy and security, with practical use cases already including flight changes, restaurant reservations, and multi-agent coordination.
These notes were generated by AI and may contain inaccuracies.
AI has reached an inflection point in the last three weeks with the release of a handful of AI assistants that are fundamentally different from typical chatbots like ChatGPT or Claude. These new assistants represent a shift beyond simple question-answering to proactive, personalized AI agents.
Instinct is a Meta product that operates through iMessage rather than requiring a separate app download. Users receive a phone number to text with, similar to WeChat functionality in China. Muse is Meta's competing product that has an app interface and is adding phone number functionality. Grockbot represents Elon Musk's project in this space.
Instinct connects to users' Google accounts including Gmail and Google Drive through a link-based authorization process. The system proactively monitors user activity and learns from responses to understand priorities and attention-worthy items. The company behind Instinct is a startup that achieved a $10 billion valuation within four months of launch.
Trust considerations favor established companies like Meta over smaller startups for privacy and access reasons. Meta provides more battle-tested security and has conducted briefings on their security model for private instances. Established companies offer additional checks and balances compared to rapidly scaling startups.
Instinct demonstrated proactive assistance by notifying users about prescription readiness and providing timing recommendations based on flight schedules. The system suggested optimal pickup times considering airport departure requirements.
Three months prior, a book preface was drafted discussing the evolution from full-time in-person assistants to virtual assistants to digitally enhanced virtual assistants. Early reader feedback suggested these capabilities addressed "1% problems," but the technology has rapidly advanced to handle 99% of described use cases.
Instinct has successfully changed flights, secured hard-to-get restaurant reservations through OpenTable by monitoring drop times, and coordinated schedules between multiple users. When multiple friends have Instinct, their AI agents can become "friends" to coordinate calendar invites and scheduling autonomously.
ChatGPT allows memory exports, but the goal for these platforms is creating durable storage of user life, story, history, and preferences. Agentic AI systems require extensive authorization of personal data, making users reluctant to repeat this process with each new model release. This creates platform lock-in as users build comprehensive personal AI relationships.
Amazon blocked Meta's agent due to concerns about circumventing sponsored placement and advertising revenue. Shopify publicly welcomed agents with an "open for business" stance. This reflects broader tension between destination websites and AI-mediated commerce where the AI becomes the entry point and can swap vendors based on price and availability.
Blackmagic's Da Vinci Resolve launched an MCP server enabling direct AI integration for video editing. Users can provide raw footage and audio to ChatGPT, which connects directly to perform editing tasks. Similar integrations now exist with Photoshop and CAD software, enabling headless operation where users never need to interact with the software interface directly.
AI interfaces eliminate the need for destination sites like Amazon, as commerce decisions happen within the AI conversation. Google previously pushed users to final destinations but didn't conduct e-commerce itself. Now AI systems handle the entire transaction flow within a single interface.
Amazon faces pressure to compete in an AI-mediated commerce environment despite advantages in economies of scale, pricing, distribution, and delivery speed. Apple Intelligence models are expected to improve significantly within 1.5-3 years. Android's Gemini will power device-level AI interactions. Multiple AI platforms including ChatGPT and Anthropic will serve as internet entry points.
Commerce will transition to conversational AI interactions where users request items and receive options comparing price, delivery speed, and vendor. This mirrors the historical shift from offline to online commerce that led to traditional retailers' decline.
ChatGPT's advanced features are available immediately to paid subscribers. Instinct remains invite-only with gradual rollout. Muse reached number one in the app store and offers easy signup. Grok's models lag behind competitors and require more technical setup. Users can call their AI agents directly without consuming AI credits, enabling hands-free interaction during activities like driving.
Keep Tim Ferriss in your library
Save the videos and channels worth coming back to, and find them again in one place.
