Podcast Summary: The Edge Has Shifted | Matt Reustle on How the Best Investors Use AI

Podcast: Excess Returns
Date: February 25, 2026
Hosts: Matt Sigler
Guest: Matt Reustle (Colossus, Business Breakdowns)

Episode Overview

This episode explores how the rapid evolution of artificial intelligence — particularly large language models (LLMs) and agentic AI — is fundamentally transforming the work of investors, allocators, and advisors. Matt Reustle shares his experiences, frameworks, and toolkits, highlighting where AI fits in the investment workflow, how to harness its power for research, due diligence, and monitoring, and what edge still remains for professionals in an AI-enhanced landscape.

Key Discussion Points & Insights

1. The Evolution of AI in Investment Research

• Early Days of LLMs:
  • The first wave of tools like ChatGPT felt more like advanced search engines or social novelties, limited in professional utility.
  • Many professionals tried and abandoned them, only to return as capabilities rapidly improved.
  • Inflection Point: Late 2024–early 2025, when "deep research models" and premium agentic AI tools enabled more complex, workflow-specific tasks.
  • “Where the big inflection came was when deep research models started to come out... agentic AI more than a traditional LLM.” [03:50] — Matt Reustle

2. LLMs vs. Agentic Workflows

• LLM (Intern Level):
  • “Calculator” analogy: handles direct, one-shot queries, returns basic, surface-level answers.
  • “The LLM in a basic free tier sense is like a calculator... like a Google search.” [05:38] — Matt Reustle
• Agentic AI (Junior Analyst Level):
  • Uses reasoning, can identify multi-layered aspects of a query, synthesizes information across sources, more like work done by a junior analyst.
  • “Agentic workflow... using reasoning... looks more like what a junior analyst or associate would do than an intern.” [05:38] — Matt Reustle

3. Where AI Fits in the Investment Workflow

• Monitoring Existing Positions:
  • Automated KPI and news monitoring, custom alerts replacing generic Bloomberg updates.
  • “I can set something up where I do that research after earnings season ends, or... it's monitoring those on an ongoing basis...” [07:44] — Matt Reustle
• Due Diligence:
  • Quick synthesis of sell-side primers, report generation, and screening for the most relevant information.
  • “There's a productivity uplift which is very visible immediately today.” [09:42] — Matt Reustle

4. Early Adoption and Leverage Points

• Real-Time Monitoring:
  • Still early days for continuous, effective real-time monitoring—but high leverage for those who set up trustworthy systems.
  • “I still think it is fairly early on that... Where... it's definitely gotten better is you can get things digested which you couldn't previously.” [12:22] — Matt Reustle
• Idea-Killing/Screening:
  • AI can help quickly flag dealbreakers (accounting irregularities, management comp issues, etc.) and reduce wasted time.
  • “We can simultaneously screen for those things but also look for things that I know are deal killers and highlight them to me...” [20:25] — Matt Reustle

5. The Importance of Prompting, Customization, and Verification

• Prompt Engineering:
  • Quality and specificity of prompts directly impacts usefulness of AI outputs.
  • Ongoing iteration and review are necessary; experiment with models, compare outputs, and adjust prompts.
  • “It really showcased to me, oh, wow, there's a lot more that I'm not tapping into because I'm not giving it the specificity that it needs.” [30:56] — Matt Reustle
  • “The best people that I know at using AI have very specific systems... and compare that to the previous model.” [33:43] — Matt Reustle
• Requesting Sources and Double-Checking Outputs:
  • Always ask for source attribution and be ready to “trust but verify.”
  • “In every prompt... ask for the sources... That way you can check through and scan to make sure that it aligns.” [22:44] — Matt Reustle

6. The AI Tech Stack

• Notable tools in Matt’s workflow:
  • Claude: Top choice for writing, editing, reframing, tone adjustments.
  • ChatGPT: Universal plugin for everything generic and deep research.
  • Gemini: For deep research, although lately some preference for ChatGPT.
  • DIA (browser AI): Embedded assistant for spreadsheets, email, and browser tasks (like Copilot for Microsoft).
  • Specialized tools: Portrait Analytics, AlphaSense for investor-specific workflows.
  • “I'm not monogamous in any way in my AI usage... For writing, Claude... ChatGPT... Gemini for deep research, I’ve gone back and forth...” [26:29] — Matt Reustle

7. Encoding Mental Models and Organizational Adoption

• Your Unique Process Matters:
  • “If I can’t articulate my mental model, a model can’t act on it.”
  • The edge lies in encoding your own frameworks, heuristics, and unique decision points into AI systems.
• Advice for Funds/Managers:
  • Start with individuals experimenting and sharing use cases rather than top-down mandates.
  • “You need to really have people experimenting at an individual level first, sharing use cases. We all do our workflows differently.” [46:53] — Matt Reustle
  • Off-the-shelf tools preferred unless you have the scale and resources to build.

Notable Quotes & Moments

• On Agentic AI vs. LLMs:
  “Agentic workflow... using reasoning to try to get at the heart of what you’re asking... it’s much more in depth on the topic using a much wider variety of sources. It’s going to give you something you can actually use that looks more like what a junior analyst or associate would do than an intern.” [05:38] — Matt Reustle

• On Prompt Quality:
  "The power of quality prompting was something that I underestimated initially... there's a lot more that I'm not tapping into because I'm not giving it the specificity that it needs." [30:56] — Matt Reustle

• On Real-Time Monitoring:
  "Are we as early to that as it feels like for me, or is this way more pervasive?" [11:47] — Matt Sigler
  "I still think it is fairly early on that in terms of who's doing that in an effective way.” [12:22] — Matt Reustle

• On the Future—Generalist vs. Specialist:
  “This is going to make generalists that have a good sense of investment decision making much more valuable... Now it doesn’t close the gap to your dedicated energy analyst, but I think it shrinks the gap quite a bit.” [53:48] — Matt Reustle

• On FOMO and the Existential Fear of Not Using AI:
  “If you’re not using these tools in some way, you’re probably a lot less productive than you could be... Over time it’s going to be like using a Bloomberg or using a calculator...” [51:22] — Matt Reustle

• On Customization as the Next Big Shift:
  "Everything is just going to be extremely customized to however we want it to be... agents will have the capacity to do a lot of things, present a lot of things, be created for you in a way that can serve you in an almost human-like assistant form." [58:10] — Matt Reustle

Techniques & Practical Takeaways

How to Start Integrating AI in Your Process

• Self-Audit:
  • Journal your research process to identify time-intensive, repeatable tasks.
• Upgrade Your Prompts:
  • Be explicit: Define role, audience, context, specific outputs, and desired sources.
  • Start simple, add detail incrementally, and always experiment.
• Use Off-the-Shelf Tools First:
  • Specialized investor tools (e.g., Portrait, AlphaSense) will handle most needs with less setup.
• Leverage Individual Experimentation:
  • Share what works within your team—collective intelligence for adoption.

Prompt Structure Example (for Deep Research/Idea Generation)

• Define role: “You are a junior analyst on an investment team at a long only fund seeking undervalued equities...”
• Define audience: “The output is for an investment committee...”
• Set scope: “Please focus on financials, especially [key KPI], provide both company and alternative source perspectives.”
• Output format: “Summarize in two pages, with bullet points of conclusions.”
• Request sources: “Include sources for all significant data points.”
  ([36:50] — Matt Reustle)

Ongoing Workflow and Strategic Mindsets

• Control & Compare Experiments:
  • Run prompts across new model releases, compare outputs, refine.
• Trust but Verify:
  • Always check attributed sources, remain skeptical of early “good” outputs.
• Iterate Frequently:
  • The most effective users regularly update and enhance their system and prompts.
• Customization is Key:
  • Your mental models, heuristics, and process steps should be reflected in your AI setup.

The Future of Investing with AI (5-Year View)

• Drastically increased customization and individualization of research and monitoring tools.
• Continuous expansion in breadth (generalists gain leverage) and compression of time needed for due diligence.
• Human judgment in interpretation and decision-making remains the core differentiator, but process automation and insight extraction accelerate dramatically.
• Early adopters gain an efficiency advantage, while industry laggards risk falling behind as productivity gains compound.

“It’s really going to open up our eyes to how much better it can be when things are individualized and made for you.”
[58:10] — Matt Reustle

Resource Links & Where to Find Matt Reustle

• Twitter: @russellmatt
• Website: matt.russell.com
• Business Breakdowns Podcast

(For full workflow and advanced prompt examples, listen from [36:29] to [41:35]. For organizational adoption, see [46:53]. For future outlook, see [57:48].)

This summary aims to spotlight the practical, tactical, and strategic lessons from one of the leading finance and AI thinkers—making it a must-listen (or must-read) for investors aiming to build or keep their edge.