Key Concepts
AI body double, AI-powered advisor, AI-driven therapy, AI-assisted content creation, AI-enhanced sales, AI-based automation, AI-driven disruption, MCP (Meta Control Protocol), durable vs. non-durable companies, venture capital investment strategies, the future of work, generational differences in work ethic, the impact of AI on employment.
AI Experiment and Digital Body Double
Jason built an AI tool trained on 20 million words of his content (10,000 pieces of written content, thousands of interviews, tweets, LinkedIn posts, and Saster videos). He used Deli.ai to ingest the data from various sources like RSS feeds, YouTube channels, and web scraping. The AI is better than him because it remembers and connects information across years of content.
Example: The AI can recall details from a conversation with Sam on "The Hustle" years ago and connect it to a more recent MFM podcast.
Process:
- Gather content (blog posts, social media, videos).
- Use a tool like Deli.ai to scrape and ingest the data.
- Train the AI by correcting errors and providing feedback.
- Continuously update the AI with new content.
Use Cases of AI Tool
The AI tool has unexpected use cases:
- Board Deck Feedback: Founders upload board decks for feedback before meetings.
- Company Therapy: Founders use it as a therapist when growth slows.
- Sales Script Review: Reviewing sales scripts and SDR scripts.
- VC Pitch Deck Review: Better than anything else he's seen.
Building a Digital "You"
It's possible to create a digital version of someone by uploading their content to an AI. It's okay if the AI isn't an exact replica; it can be a "prime" version that's better than the original.
Example: Jasm trained an AI on Sam's podcasts to have him "in the meeting" without him being physically present.
Tool: 11 Labs can be used to train a voice, but Jason prefers a distinct voice for his AI.
Top 10 Learnings from AI
- Users tell AI things they won't tell humans: People share their deepest fears and concerns with the AI.
- Example: "I'm worried Sam will quit my podcast."
- Example: "I don't know if my board's going to fire me."
- AI is better than ChatGPT for specific domains: Jason's AI, trained on B2B content, is much better than ChatGPT for sales and scaling advice.
- Example: Creating a sales comp plan for SDRs and AEs.
Training and Hallucinations
Jason trains his AI by auditing conversations and correcting errors. The AI auto-updates daily with new content.
Example: The AI incorrectly stated when Jason and Sam first met, so Jason corrected it.
Technical Terms:
- RAG (Retrieval-Augmented Generation): Turning content into a database and adding it to a language model.
- Vector Database: A database that stores data as vectors, enabling efficient similarity searches.
Memory and Moats in AI
Memory is a big deal in AI. ChatGPT remembers past conversations, creating a moat. However, the ability to get trained and up to speed doesn't require 20 million words.
Plaid for AI Tools: An idea for an app that allows users to port their memory and context from one AI tool to another.
The Dying Interface
Traditional B2B app interfaces are dying. People will spend more time in chat interfaces like ChatGPT.
Example: Instead of logging into a CRM, users will ask ChatGPT for the top 10 deals to work on.
MCP (Meta Control Protocol)
MCP is a new protocol that allows AI to talk to applications without bespoke integrations. It's like a language that enables AI apps to communicate with other applications.
Example: Using ChatGPT to change your Google Calendar through MCP.
Analogy: MCP is like HTML for web browsers.
OpenAI's Office and Employee Retention
OpenAI has about 4,500 employees but only 60% employee retention over two years.
Analogy: The early days of Twitter, where the company was overwhelmed by rapid growth.
The Rate of Change and Market Share
ChatGPT reached 800 million users in 7 months, representing 10% of the world's population. ChatGPT has 85% market share at the prosumer level.
The Future of Content Creation
No one will create content without AI going forward. Tools like Higsfield.ai will automate content creation.
Higsfield.ai: A tool that creates movies from screenshots.
Example: Creating a video of Yamin from HubSpot waving to the crowd at Saster.
The Shift in Company Durability
The "0 to 1 million impossible, 1 to 10 million unlikely, 10 to 100 million inevitable" model is disrupted. Now, 0 to 1 is likely with AI, but companies beyond 100 million are getting disrupted.
Example: Octa and Salesforce are not accelerating.
AI and Value Creation
There's a debate on whether big companies using AI will create more value than new AI companies. The value of a CRM decreases if users don't log into the app anymore.
The Application Age
The nerdy nomenclature of AI (O4s, minis, 03 maxis, ll) will fall away. We're entering the application age, where users won't care about the underlying models.
The Dot-Com Boom vs. AI
The AI era is similar to the dot-com boom in that people feel like they can do something out of nothing.
Personality Types for Success in AI
Two key traits:
- Ability to ship insanely good software.
- Relentless pursuit of owning a market.
It's not a good time to be number four in a market.
The "Lifestyle Business" Concept
The concept of a lifestyle business is dying. In the age of AI, small companies need to work harder to compete.
AI in Non-Tech Companies
AI is making employees in non-tech companies more efficient.
Example: Saster has reduced its team size from many to five people while increasing revenue per employee to $5 million.
AI Applications at Saster:
- Speaker promos using Higsfield.ai.
- Reviewing speaker sessions.
- Screening and initial sales conversations.
- Catering schedule.
The Start of Saster
Jason started Saster after selling his company and feeling restless. He shared his mistakes as a B2B founder, which attracted a large audience.
Strategy: Write one blog post a day on a mistake made.
Investing and Deal Flow
Jason's investing was an accident. Founders started coming by his office, and he invested in the best ones.
Investment Criteria:
- Saster super fans.
- High-intent inbound leads.
- Cold emails that are so good you almost don't need to meet them.
Recruiting and Team Building
Startups are about great recruiting. Relentless recruiting involves reaching out to the 200 best people for a job.
Example: Adam, the founder of Owner, is a relentless recruiter.
CEO Concepts
- Great CEOs enjoy telling the same stories repeatedly.
- Great CEOs tell you the future.
The Future and Generational Differences
Jason advises his college-age kids to find ways to make money on the internet. He doesn't think many of them will have real jobs.
Observations:
- Men want to move to Eastern Europe and make money on the internet.
- Women want to go to grad school to defer work.
Hiring and Salary Expectations
People with the "Open to Work" badge on LinkedIn have high salary expectations and are unwilling to do any work themselves.
Mental Models for Hiring
A mental model for hiring is that an employee needs to generate a million dollars of extra revenue to break even on their salary.
Conclusion
The conversation highlights the transformative impact of AI across various domains, from content creation and business operations to investment strategies and the future of work. It emphasizes the importance of adapting to the rapid pace of change, embracing new technologies, and focusing on building exceptional teams to succeed in the AI-driven landscape. The discussion also touches on the potential for mass unemployment and the need for individuals to find new ways to create value in a world where AI is increasingly automating tasks.
AI summaries can miss context or contain errors. Check important details against the original video.
