“We raised $22M — to reinvent the crypto trading frontend in the form of an agentic browser.” A while back at the Silicon Valley 101 conference in San Jose, I shared our insights on agents and crypto trading, envisioning the next-gen agentic trading frontend. After the talk, the crowd was packed—I couldn't move: "What are the pain points in crypto trading right now?" "Why an agentic browser?" "Why on-chain trading?" Here are some of my thoughts: (1/5)

Crypto trading today feels more like an arena for institutions and pros. Retail traders are often at a disadvantage: info overload, delayed executions, emotional decisions, and orders getting front-run by market makers. Even staring at screens 18 hours a day doesn't cover 24/7. To flip the script, we need to automate the "discovery → decision → execution" loop, with every step verifiable and traceable. (2/5)

Many still stick to centralized exchanges, but on-chain opportunities aren't just real—they're exploding. Just look at the numbers: On-chain spot daily volume exceeds $20B, and on-chain perps have hit trillions this year, several times higher than at the start. The on-chain world is more complex—more steps, strategies, and fragmented info—but that's exactly why AI can deliver outsized efficiency gains, directly impacting results. (3/5)

That's why we're building an agentic browser. The browser is already your single entry point for info and orders; unifying research and execution turns "spotting opportunities" into seamless "matched trades." LLM advancements make extracting signals from messy data feasible; automated code gen and quant modeling lower the bar for strategy dev and iteration. Breakthroughs in Donut for data scraping, unstructured understanding, and multi-dimensional analysis ensure stable signal outputs on the research side, then land them via strategy routing, risk control, and timing on execution. (4/5)

Connected with top AI practitioners, scholars, and hedge funds in Silicon Valley. Turns out many traditional hedge funds are building internal trading agents too—they're just more closed-off. Our approach: Bring this full stack on-chain, into the browser, making speed, discipline, and verifiability weapons every trader can wield. (5/5)

The potential of AI has yet to be fully unleashed. The intelligence behind large models has not yet been transformed into a capability that everyone can use in their daily lives. Everyone deserves an AI quantitative analyst who truly understands them, always by their side. Evolving from multiple fragmented, passive interfaces into a proactive, seamlessly connected, and truly understanding intelligent companion. [Final]