How to Raise Your Agent Episode 8 | Matthew Cheung, CEO, ipushpull

In the OTC markets that make up roughly half of global trading volume, people still type prices into chat windows with their fingers, watch responses with their eyeballs, and scribble numbers on scratch paper. Matthew Cheung, CEO of ipushpull, has spent two decades building communications infrastructure for financial markets, and he argues that this messy, unstructured chat data is about to become the most valuable untapped resource in finance — unlocked by the convergence of large language models, AI agents, and a governance framework that treats compliance not as a burden but as a performance engine.

Why chat won the interface war

Cheung sees the evolution of human-computer interaction in financial services as a clear progression: graphical user interfaces in the 1990s, APIs in the 2000s, and now agents as the next interface moment. But the underlying channel that connects all of them is chat. More than half the world's population uses chat platforms, from WhatsApp's three billion users to Microsoft Teams inside enterprises. Chat has been around for 50 years, but it was always designed for human-to-human communication. The problem was that the data inside those conversations — the prices, the axes, the market color, the relationship-building chit-chat — was completely unstructured and effectively inaccessible.