Implied CEO Ying Hua joined Brett Caughran and Khe Hy on Fundamental Edge to talk about why public equities are one of the hardest domains for AI, and why that difficulty is what convinced her to start Implied. In public markets, what matters most to a stock can change with every earnings release or news event. The context needed to make sense of those changes is scattered across sources and pieced together differently by every analyst.
"When you read the surface level reports, they will have one interpretation. But when you have more domain knowledge, depth of domain knowledge, you're going to reach a different conclusion."
Her test for whether a platform has any of that depth: ask it to summarize JP Morgan earnings. If revenue and margin come back on top, the domain knowledge isn't there.
Much of the episode digs into other problems we work on every day, including:
- Why Implied runs its own live transcription instead of buying clean transcripts, and the value of raw data for sentiment analysis
- How "just point Claude Code at it" breaks down in production: data updating at random intervals, scrapers that break when a website changes, and no way for a non-programmer to catch a logic error in code
- Why an agent driving Excel through Python hits a ceiling, and how a cloud-based, AI-native spreadsheet solves it
- Why judgment stays with the analyst, and what an objective, verifiable, and constantly updating knowledge layer looks like
You can also watch the episode on YouTube.