Using AI to Find Investing Stories with Perscient Co-Founder Ben Hunt

What It Covers

Ben Hunt, cofounder of Persient, explains how his company uses AI to process 200+ billion tokens of global media to identify emerging market narratives before they become mainstream, tracking semantic patterns across thousands of stories in multiple languages.

Key Questions Answered

• •Narrative Machine Technology: Persient processes everything published globally—newspapers, transcripts, websites in all languages—using large language models constrained by human-directed context engineering. The system achieved 100x improvement in signal detection by transitioning from handcrafted small language models to AI-powered semantic signature tracking across multiple languages.
• •Context Engineering Over Prompting: Successful AI implementation requires controlling all inputs and outputs, not asking open-ended questions. Users must provide the data, define allowed thought processes, and validate results through secondary AI systems. This constraint prevents hallucination and ensures consistent, reproducible answers for the same inputs.
• •Discovery Phase Alpha Generation: Value investors make the most money during the discovery phase when markets wake up to undervalued assets. Persient tracks when dormant narratives resurface in media, identifying approximately twelve core stories per sector that repeat with different company names, allowing investors to time entry before widespread recognition.
• •Forward Guidance as Primary Policy Tool: Central banks shifted from traditional monetary policy to using coordinated communication as their main toolkit starting in 2009. Bernanke admitted in his valedictory address that forward guidance worked better than quantitative easing, fundamentally changing how markets respond to policy and making narrative tracking essential.
• •Immigration Narrative Reversal: Data shows significant decline in anti-immigration sentiment and increase in pro-immigration stories across all political affiliations since October 2024, contradicting policy direction and social media perception. This disconnect between actual public opinion measured through comprehensive media analysis versus perceived sentiment suggests potential political realignment in upcoming midterm elections.