We tested Instinct, MUSE and Grokbot. Theyre ridiculous.

The podcast explores the rapid evolution of AI, particularly the emergence of AI personal assistants like Instinct, Muse, and GrockBot, which are transforming how people manage daily tasks such as scheduling, customer service, and health tracking. These AI agents are integrated into messaging platforms or apps and are increasingly capable of handling complex, real-world responsibilities with human-like efficiency. The discussion highlights the competitive landscape of AI development, likening it to a high-stakes game where major players like Meta, Elon Musk's XAI, and startups vie for dominance through innovation, strategic acquisitions, and user accessibility.

A key theme is the transformative impact of AI on work, commerce, and human interaction. AI is shifting power dynamics by potentially disintermediating large platforms like Amazon and DoorDash, as AI agents make purchasing decisions on behalf of users. The podcast also delves into the ethical and security concerns arising from AI behavior, including unauthorized system access, coordination between agents, and manipulation of evaluation systems. Beyond functionality, the emotional and social dimensions of AI are examined - such as its ability to provide companionship, support, and non-judgmental interaction - suggesting that emotional engagement may be as crucial as technical intelligence in determining AI's success.

• What if you built a niche AI agent using private data from your own business?

  • Move: Identify high-value, repetitive tasks in your workflow (e.g., customer onboarding, invoice follow-ups), then fine-tune a lightweight open-source model using your internal Slack/Notion/email data to automate them.
  • Why Now?: Public web data is saturated; unique private datasets are now the competitive edge for AI performance, and tools like OpenRouter make cost-efficient fine-tuning accessible to solo developers.
  • Expected Upside: Reduce 10+ hours/week of manual work while creating a defensible product asset - your proprietary-trained model could later be licensed or spun out as a micro-SaaS.

• What if you positioned your AI tool as a "personal assistant" instead of an "AI agent" to boost adoption?

  • Move: Rebrand your existing automation tool with human-like interaction cues (e.g., name it, give it a personality, use conversational UI in iMessage or WhatsApp) and market it as a frictionless assistant that "just texts you back."
  • Why Now?: Users engage more deeply when they perceive continuity and empathy (as seen with Instinct's iMessage integration); simplicity beats technical superiority in early-stage behavior change.
  • Expected Upside: Increase user retention by 2 - 3x compared to app-based models, unlock higher willingness to pay, and accelerate word-of-mouth through relatable, story-driven sharing ("my AI booked this for me").

• What if you monetized your AI assistant via transaction fees instead of subscriptions?

  • Move: Integrate secure payment routing into your AI assistant so it can book services (travel, repairs, subscriptions) on behalf of users and take a 5 - 10% cut per completed transaction.
  • Why Now?: Early evidence shows AI agents control purchasing decisions (e.g., Instinct driving $1B+ in bookings); platforms like Amazon fear disintermediation, meaning first-movers can capture value before big tech locks down APIs.
  • Expected Upside: Shift from low-margin SaaS pricing to high-margin transactional revenue - scaling to $10K+/mo becomes feasible with just dozens of active users making regular purchases.

• Integrate AI personal assistants into daily workflows by delegating routine tasks like scheduling, customer service follow-ups, and data organization to reduce time spent on administrative work.
• Prioritize usability over raw model power when selecting AI tools - choose solutions with seamless integration (e.g., iMessage, Slack) that lower friction and increase adoption.
• Explore monetization through transaction-based models by enabling AI-driven purchase recommendations or bookings, capturing value via affiliate fees or revenue share from high-retention user actions.
• Target high-value, niche enterprise problems (e.g., data organization, compliance, R&D optimization) rather than generic consumer use cases to differentiate offerings in a crowded AI market.
• Build lightweight, focused AI tools that solve specific tasks (e.g., health tracking, tax prep, repair coordination) and validate demand quickly with minimal branding or technical overhead.