Podcast and Takeaways: Your ‘Move 37’ Moment: When AI Sees What You Don’t
Jun 9, 2026
/
By Sean Bailey, Horsesmouth Editor in Chief
AI for Advisors Podcast: In 2016, a machine made a move no human would ever play, and it changed the game forever. Now AI is passing the CFA Level III exam and acing personal finance questions it got completely wrong just two years ago. Learn why every advisor is sitting in front of a ‘Move 37’ moment of their own, and the three moves of AI fluency: observe, experiment, and investigate, with ready-to-use prompts for each.
Key takeaways
- In 2016, AlphaGo’s “Move 37” against Go world champion Lee Sedol stunned experts; the move that looked like a mistake set up an attack dozens of moves ahead. It was the instant machines stopped being mere calculators and started being creators.
- NYU Stern researchers tested 21 large language models on the CFA Level III exam, the most judgment-heavy test in finance. Top models scored 79% (ChatGPT) and 76% (Gemini) against a 63% passing bar.
- In January 2023, a Claremont College economist tested ChatGPT on 14 common personal finance questions. It answered eloquently but totally wrong. When Horsesmouth reran the same questions on today’s models, every answer was correct and nuanced.
- Move 1, observe: Find your hidden time drains. Try this prompt: “Based on what you’ve learned from our recent chats, identify five ways I could improve the way I’m using ChatGPT and running my advisory business, whether through smarter prompts, better systems, or non-AI efficiency fixes.”
- Move 2, experiment: Give AI one real task a day, like drafting a 200-word market update for anxious clients, summarizing meeting notes (with no personally identifiable information) into three action items and a follow-up email, or generating LinkedIn post ideas.
- Move 3, investigate: Use deep research, the semi-autonomous agent inside ChatGPT, Gemini, Claude, and Copilot that reviews dozens or even hundreds of sources and returns a synthesized, cited briefing.
- Structure deep research prompts with role, task, format, and context, the core of the Horsesmouth power prompt framework. Context is the who, what, when, where, and why of your request.
- Deep research is augmentation, not automation. It works like a research partner, not a summer intern. The question isn’t whether AI can replace advisors; it’s what kind of advisor you become once you start leading AI.