They Just Invented the Agentic AI Brokerage

Here's a comprehensive summary of the YouTube video transcript:

Key Concepts

• Generated Assets: A feature on Public.com that allows users to create custom index funds by providing a single text prompt with their investment thesis.
• Agentic Brokerage: An AI-powered agent mode for the Public.com platform that can execute complex trading instructions based on natural language prompts.
• Custom Indexing: The ability to create personalized investment portfolios that function like index funds but are tailored to specific user-defined criteria.
• Fractional Trading: The capability to buy and sell portions of shares, enabling investment in custom indexes with many holdings without requiring large capital.
• Asset Score: A metric used by Public.com to evaluate the performance of generated assets, considering both returns and risk (drawdowns).

Generated Assets: Creating Your Own Index Funds

Public.com has launched "Generated Assets," a feature that allows users to create their own index funds through a simple text prompt. This innovation is described as a "lovable for financial products," enabling users to articulate an investment thesis and have the platform assemble a portfolio that aligns with it.

Key Features and Functionality:

• Prompt-Based Creation: Users enter a single prompt detailing their investment idea (e.g., "build us an index of utilities with the highest earnings growth rate expected for 2026").
• Stock Evaluation: The AI evaluates all stocks in the public markets based on the prompt's criteria.
• Customizable Weighting: Users can specify how the index should be weighted, including equal weighting, weighting by relevancy score (AI-determined importance), or market cap weighting.
• Single Asset Investment: Once created, the custom index appears as a single asset on the user's portfolio, allowing for 100% investment with a single click.
• Eliminates Manual Research: This feature removes the need for users to research dozens of tickers, decide on inclusions/exclusions, or manually calculate weightings, thus eliminating the need for spreadsheets.
• Editability: Users can still edit and customize the generated index further.

Examples of Prompts and Use Cases:

• Thematic Investing:
  • "Business models with low exposure to AI."
  • "Companies with high R&D spending."
  • "Companies with Andreessen Horowitz invested in."
  • "Invest in GLP1 drugs."
  • "Self-driving automation, B2B companies with less than 10%."
• Fundamental/Metric-Based:
  • Companies with low average revenue per user but a large user base (potential for revenue growth without marketing spend).
  • Companies that have aggressively reduced debt in the recent 12 months.
• Qualitative Investing:
  • Companies that are "relentless."
  • Companies that "over-promise and over-deliver on earnings calls."
  • Companies with "aggressive CEOs."
• ETF Remixing:
  • "Give me the S&P 100, but take out companies in AI or exposed to tariffs."
  • "ARC minus Tesla" (remixing a celebrity ETF).
• Personalized Themes:
  • "Companies that benefit from Gen Z consumer habits and preferences, including digital-first brands, social media platforms, sustainable products, and experiential retail."
  • Combining multiple interests like music, race car driving, and vegan lifestyle into a single index.

Underlying AI Capabilities:

• Master Agent: Interprets user intent.
• Thousands of Underlying Agents: Evaluate individual stocks based on interpreted criteria.
• Mathematical Ranking: Surfaces results even for qualitative criteria.
• Sentiment Analysis: Incorporates opinions from articles, press, and analyst reports.

Backtesting and Asset Score:

Generated assets undergo a full backtest against the S&P 500, providing data on returns and drawdowns. The "Asset Score" is a proprietary metric that evaluates performance, considering stability and diversification, with a score above one indicating outperformance relative to the S&P 500's Sharpe ratio.

Agentic Brokerage: AI-Powered Trading Execution

The second major innovation is the "Agentic Brokerage," which introduces an agent mode to the Public.com platform. This AI agent can execute complex trading instructions based on natural language prompts, mimicking the capabilities of a human financial advisor.

Key Features and Functionality:

• Natural Language Instructions: Users can instruct the agent with specific goals and constraints (e.g., "I need $50,000 for a down payment next month. Only sell things at long-term capital gains, trading close to all-time highs, with a PE ratio of X").
• Automated Portfolio Management: The agent can analyze a user's portfolio across hundreds of positions, schedule sales, and execute them.
• Complex Scenarios: Handles instructions like:
  • "If the Fed cuts rates, trim 10% of my bank stocks and move it into high-growth tech."
  • "Set up an automated cash sweep every time I have more than $5,000 in my Chase checking into the Public bond account."
  • "Vote yes to Elon's compensation package."
• Intent Understanding: The agent aims to understand the user's intent, not just the literal instruction.
• Efficiency: Significantly speeds up processes that would traditionally require phone calls to a brokerage and lengthy customer service interactions.
• Accessibility: Provides a level of service typically reserved for wealth managers to a broader audience.

Comparison to Traditional Brokerages:

• Traditional brokerages often have outdated technology, requiring users to navigate multiple screens and hunt for specific settings (e.g., reinvesting dividends ticker by ticker).
• Agentic Brokerage offers a more intuitive, conversational interface, akin to a user interface upgrade from web to mobile, now to natural language.
• It addresses the capacity limitations of traditional brokerages, especially during high-volume market moments.

Rollout and Availability:

• Agentic Brokerage is slated for rollout in Q1, with a waitlist available on Public.com. Access will be rolled out starting in January.

Broader Implications and Future Vision

• Democratization of Financial Services: Both features aim to provide sophisticated investment tools and services to a wider range of investors.
• Customization as a Trend: The emphasis on customization aligns with a broader trend across various industries.
• Shift from Beta to Alpha: With the return of higher interest rates, there's a renewed focus on generating alpha (outperformance) beyond just market beta.
• Addressing S&P 500 Concentration: The rise of highly concentrated indexes (e.g., 40% of the S&P 500 in 10 stocks) is driving demand for alternative investment strategies.
• Generational Shift: Younger, digitally native investors (millennials) are more confident in their financial ideas and abilities, fueled by online communities and increased financial literacy.
• Future Expansion: Generated Assets is expected to expand to include options, cryptocurrency, and bonds, potentially leading to more structured products.
• Community and Sharing: A "share" button for generated assets is coming soon, allowing users to share their creations with the community and potentially build followings as "generators."
• Comparison to Motif: The concept of user-created investment themes is reminiscent of Motif, but Public.com's fractional trading engine and AI capabilities offer a more advanced and scalable solution.

Live Demo and User Experience

The transcript includes a live demonstration of the "Generated Assets" feature.

Demo Walkthrough:

1. Prompt Input: A prompt is entered: "High user base, low average revenue per user."
2. Evaluation Process: The system dispatches "evaluation agents" to research individual stocks based on the criteria.
3. Stock Population: Stocks begin to populate the generated index, ranked by relevancy score.
4. Index Creation: The AI identifies 25 companies, including PayPal, Starbucks, Zoom, Spotify, and eBay.
5. Weighting: The index is equally weighted as no specific weighting preference was given.
6. Company Analysis: Clicking on a company (e.g., PayPal) reveals its relevancy score and the AI's reasoning, including calculations of revenue per user.
7. Backtesting and Asset Score: The generated index is backtested against the S&P 500, showing superior returns and drawdowns. The Asset Score is presented as a more digestible metric than just raw returns.
8. Second Prompt Example: A prompt for "Companies that benefit from Gen Z consumer habits" is demonstrated, yielding companies like Spotify, Pinterest, Airbnb, Shopify, Live Nation, Netflix, Uber, and Starbucks.
9. Detailed Explanations: Clicking on individual companies provides specific reasons for their inclusion, such as Starbucks' relevancy score and PlayStation's benefit from Gen Z gaming habits for Sony.
10. Multi-Theme Prompt: A prompt combining multiple interests (music, race car driving, vegan lifestyle) is attempted, showcasing the AI's ability to handle complex, multi-faceted requests. A minor technical glitch with commas in the prompt is encountered during the demo, highlighting the iterative nature of AI development.

Conclusion and Call to Action

The speakers express excitement about the potential of both Generated Assets and Agentic Brokerage. They emphasize that these tools empower individual investors with sophisticated capabilities previously only accessible to institutional investors or those with significant wealth.

Call to Action:

• Users are encouraged to sign up for Public.com, download the app, and log in.
• Generated Assets are available now on mobile and web.
• Agentic Brokerage will be available in Q1, with sign-ups for the waitlist encouraged.
• Funding an account is suggested to expedite access to new features.