How Refi Trading Puts Wall Street AI in Your Hands—Without Losing Control

What we talked about.

Zeeshan joins Sunny Ray to discuss Refi Trading, a non-custodial AI trading infrastructure that brings institutional grade execution to everyday traders without requiring them to give up control of their assets. Zeeshan traces his path from growing up in Edmonton, delivering newspapers, and building websites for NHL players, through decades in financial systems and emerging tech, to becoming an Nvidia and Anthropic certified instructor. He explains how a stuck international wire transfer through Barclays pushed him toward blockchain as trust infrastructure rather than speculation. The conversation covers Refi Trading's reinforcement learning agents that process hundreds of signals to rebalance portfolios within user defined risk limits, and P402, agentic payments infrastructure that routes requests across hundreds of LLMs and settles in USDC on Base. Zeeshan also reflects on his father's immigrant journey from Pakistan to Canada as a formative influence, his teaching philosophy, and his broader vision of autonomous finance where AI, crypto rails, and compliance combine to widen access to tools once reserved for hedge funds.

A non-custodial AI trading founder explains how Wall Street grade execution can reach retail traders without surrendering control.

The questions, and the answers.

What is Refi Trading in an elevator pitch?

Refi Trading is non-custodial AI trading infrastructure for people who want institutional grade execution without handing over control of their assets. Users connect their broker, define their own risk limits, and deploy our reinforcement learning driven execution within those boundaries. We use cryptographic proof gates and audit trails built into the workflow, designing from bottom up compliance. It is Wall Street grade automation made radically more accessible while the user stays in control.

What was your first real job that shaped how you think today?

I'm going to take you back to when we used to deliver newspapers. You had to get up early so you were disciplined, you had to collect subscription money, and you had to be reliable and on time. Once I had my own route figured out, I actually subcontracted it to another kid in my neighborhood and went and got an apartment block that took less time. I started understanding business at its core and how to scale it back then.

The story goes that an international wire transfer got stuck for 60 days and broke something in your head. What actually happened?

I moved from Edmonton to Dublin, then Melbourne for six years, then Dubai, then New York, and now I spend time in Georgia. I sent a wire from myself to myself and it did not come through. Barclays, an intermediary I had nothing to do with, started asking compliance questions. I realized why should we deal with third party intermediaries just to send funds to myself, so I started looking into Ethereum and digital crypto systems.

Who are some of your early mentors or influences that shaped this first principles approach?

I can start with my dad. He was so brave for coming to Canada from Pakistan in 1967 when there were hardly any immigrants from Pakistan. He and my mother came as university students and started a restaurant, even though he was studying to be a geologist. He later became a stockbroker on the old Alberta stock exchange. I used to go to his office and punch on the Bloomberg terminal, my first computer. He remains a mentor to me.

Give the 60 second pitch for Refi Trading. What is it and who is it for?

Refi Trading is non-custodial AI infrastructure for institutional grade execution without handing over control of assets. A user connects their broker, defines risk limits, and deploys a reinforcement learning agent using 380 to 400 dynamic signals to rebalance their portfolio. We never touch client money. A new SEC rule, 203A, lets companies like ours run end to end internet advisory for individual traders, tools previously locked behind hedge funds.

P402 is described as agentic payments infrastructure for the AI economy. What does that mean and why does the AI economy need its own payment rails?

I saw students building capstone projects with frontier models that worked like toys but hit a brick wall when it came time to scale, then got billed more than they could afford. P402 sits one layer below Refi as payment and routing infrastructure for AI native commerce. It routes requests across 300 plus LLM models, uses an OpenAI compatible interface, and settles with gasless USDC on Base, giving every request traceable cost, route, and settlement visibility.

What is autonomous finance, and how close are we actually to it?

Autonomous finance is a shift from humans manually coordinating every financial action to systems that sense, decide, verify, and execute within constraints. We are not fully autonomous yet and should not pretend we are, but pieces are already here: AI driven decision support, automated execution, proof based controls, and self custody infrastructure. The real work now is stitching those pieces together in a trustless way.