How AI Tools Are Rapidly Disrupting the Investment Industry | Brett Caughran & David Plon
Timestamps are as accurate as they can be but may be slightly off. We encourage you to listen to the full context.
Podcast Summary
In this episode, Brett Karin from Fundamental Edge and David Plon from Portrait Analytics dissect the AI revolution transforming hedge fund research. They explore how sophisticated tools like GPT-4 with deep research capabilities (02:42) are becoming indispensable for investment professionals, moving from skeptical experimentation to competitive advantage. The discussion covers practical implementation strategies—from prompt engineering workflows to thesis monitoring systems—while addressing critical limitations including quantitative modeling challenges (44:02) and the ongoing need for human expertise in conviction building. Both experts emphasize that successful AI adoption requires clearly defined research processes (23:34) and thoughtful integration rather than wholesale outsourcing of analytical work.
Speakers
Brett Karin
Founder and head trainer at Fundamental Edge, a buy-side analyst training platform. Former analyst at a tiger cub that made its name shorting the tech bubble and banks during the GFC, bringing a skeptical approach to AI hype while recognizing transformative tools when they emerge.
David Plon
Co-founder and CEO of AI-powered investment research platform Portrait Analytics. Former buy-side analyst at Barclays, SlatePath, and Bowpost who left the traditional analyst track in 2021 to build AI tools specifically for fundamental investment research.
Max (Host)
Host of Other People's Money podcast, conducting in-depth conversations with investment professionals about industry trends and technology adoption in institutional asset management.
Key Takeaways
Decompose Your Investment Process Before You Accelerate It
Create a detailed research document defining your 12-15 critical investment workflows step-by-step. You can't augment a broken process - understand what drives alpha in your approach before layering AI on top. (13:24) Brett warns: "if more than one third of your investment research motion is outsourced to AI, that's probably too much."
AI Makes Smart People Smarter and Dumb People Dumber
The best AI implementation requires thoughtful experimentation and domain expertise to identify what's valuable versus garbage output. (34:00) Without investment context, AI-generated research may look impressive but contain critical flaws that only experienced professionals can catch.
Ground Your Models to Eliminate Hallucinations
Pair general-purpose models (ChatGPT, Claude) with grounded tools like NotebookLM by uploading your filings, research notes, and internal documents. (11:48) This creates an "anchored" system that minimizes hallucinations while maintaining access to your proprietary research stack.
Focus on Qualitative Leverage, Avoid Quantitative Pitfalls
AI excels at language-based research tasks but fails miserably with dates, numbers, and Excel modeling. (44:05) Use AI for getting up to speed on industries, thesis monitoring, and qualitative synthesis - but never trust it with financial calculations or building three-statement models.
Build Competitive Advantage Through Agentic Workflows
Create "thesis monitoring" systems that act like junior analysts watching your coverage universe 24/7. (15:36) Set up automated alerts for key thesis inflection points - like executive compensation changes or Medicare Advantage utilization trends - to scale your research bandwidth without hiring additional analysts.
Statistics & Facts
- 25% to 30% of Brett's investment research motion can now be outsourced to AI, compared to just 2% six months ago (14:47) - marking a significant acceleration in AI adoption for investment research workflows.
- AI will make smart people smarter and dumb people dumber (33:54) - highlighting how AI effectiveness depends entirely on the user's expertise and ability to direct the technology properly.
- The average prompt in foundation models is only nine words (40:30) - demonstrating that most Wall Street professionals are "grunting into the context window" rather than utilizing proper prompting techniques for optimal AI output.
