Make Your Own Financial Agent for Investors

Agenda Overview 🌟

Section 1: Agent Architectures

• Dive into foundational concepts of agentic AI: single vs. multi-agent systems.
• Discover the ReAct (Reason-Act-Observe) loops essential for decision-making and function APIs for accessing data.
• Demonstration: Utilize a LangGraph agent to query Yahoo Finance for cap table stress-tests.
• Assignment: Personalize a GitHub template tailored to your investment sectors.
• Demystifying AI Agents: Analyze LLMs, memory, tools, and the Observe → Think → Act cycle.
• Compare various architectures — monolithic vs. multi-agent and heavyweight frameworks vs. minimalist harnesses.
• Explore pi-mono: an open-source toolkit centered around "primitives, not features" for building custom agents.
• Live Building: Code a real VC due diligence research agent from scratch in our final session!

Section 2: Automated Deal Sourcing

• Craft agents for lead generation utilizing vector embeddings on Crunchbase/LinkedIn data.
• Engage in discussions on prompt engineering to assess founder teams.
• Hands-On Integration: Merge Airtable with OpenAI agents for real-time matching; evaluate the performance using precision and recall metrics.

Section 3: Advanced Due Diligence

• Deploy hierarchical agents for in-depth financial analysis and risk management.
• Emphasize memory storage for audit trails and ensuring human oversight.
• Demo: Backtest your portfolio to identify performance enhancements; complete a diligence simulation using live data as homework.

Section 4: Portfolio Agents & Scaling

• Coordinate multi-agent systems for effectively monitoring, exit signals, and rebalancing strategies.
• Discuss cost considerations, addressing errors, and various deployment methods.
• Capstone Project: Collaboratively design a "VC Vault Agent" to bring all your learning together!

🌍 Location: Frontier Tower @ Makerspace, 995 Market Street, San Francisco
🎟️ Tickets: Get your ticket here!