Crypto Platforms Race to Deploy AI Agents

AI agents in crypto trading aren't some distant future anymore — they're the thing every major platform is quietly scrambling to build right now. Pranav Ramesh, head of options research at Nasdaq and co-founder of the AI startup Leadpoet, sat down to offer a perspective that most executives in his position would soften with caveats. He didn't. The industry is moving fast. Jobs are already going. And crypto, he argues, is going to be the sector that defines how AI agents get used in retail trading — not Wall Street, not Silicon Valley.

What Is Nasdaq Doing With AI Agents Right Now?

Ramesh said Nasdaq has been running AI agents across multiple business functions for roughly the past 18 months, with deployment accelerating sharply as model reliability improved. The areas he named: market surveillance, compliance workflows, and market microstructure analysis. The part that matters most isn't the list — it's his explanation of why adoption moved slowly at first. Earlier generations hallucinated too often. You can't have a compliance system inventing facts.

The specific product he pointed to was Nasdaq Verafin, which launched what it calls an 'Agentic AI Workforce' — a system built to automate anti-money laundering processes that are high in volume but low in judgment requirements. The pitch is straightforward: let agents handle the repetitive screening, free up human analysts for the cases that actually need thinking.

The second example he raised was harder to ignore. Nasdaq Dynamic M-ELO became the first exchange AI-powered order type ever approved by the SEC in 2023. The system pulls from more than 140 factors to adjust order behavior in real time — no human touching it between signal and execution. That's not a proof of concept. That's live infrastructure on a regulated exchange.

Crypto's Advantage Over Traditional Finance

Here's the take that Ramesh pushed hardest: crypto is going to move faster than traditional finance on AI agents for retail, and it's going to do it because the environment permits it. Fewer legacy systems to work around. Higher risk tolerance from the user base. And a culture that already treats automation as a feature rather than a threat.

He sees crypto platforms building out position analysis tools, trade suggestion engines, and execution support — all AI-driven, all aimed at the retail trader who currently relies on their own gut or a Discord server. The model he described isn't fully autonomous, though. Agents handle the bulk of analysis and workflow generation, but a human keeps the final approval. That matches what Nasdaq still does internally: systems stop short of full automation, with human review as the last gate.

"The crypto trading world is actually going to lead the charge on how AI is used within the retail trading environment," Ramesh said in the interview. Whether that's optimism or a straightforward reading of incentive structures, the argument holds up. Crypto platforms don't face the same regulatory friction that traditional brokerages do when they want to change how orders get routed or how recommendations get generated.

The crypto trading world is actually going to lead the charge on how AI is used within the retail trading environment.

Yes, the Jobs Are Going — and That's Not a Prediction

Ramesh was blunter on labor than most people in his position tend to be. "Yes, it will take a lot of jobs," he said directly. He named lower-level software roles, customer service positions, and junior analyst jobs as already being displaced — not at risk, already being displaced. He framed it as something he's observed, not something he's forecasting.

The data backing him up is hard to ignore. Crypto.com cut 12% of its staff in a push toward automation and efficiency. Messari, the crypto research firm, parted ways with multiple employees and its CEO as it repositioned itself around an AI-first operating model. And Block — Jack Dorsey's payments company — announced plans to eliminate 40% of its workforce, more than 4,000 people, with AI capability improvements cited as the driver.

That's three separate companies in crypto and adjacent finance, across a relatively short window, all pointing at the same thing. Ramesh isn't making a provocative claim. He's describing a trend that's already visible in the quarterly headcount numbers.

What Is Leadpoet and Why Does It Matter Here?

Ramesh co-founded Leadpoet with Gavin Zaentz — the two met while both working at Nasdaq. The company's origin story tracks with everything Ramesh described about his time at the exchange: he kept running into outbound sales tools that could generate lists but couldn't identify genuine buying intent without a human analyst doing manual research on top. Leadpoet is their answer to that gap.

The platform positions itself as an AI-powered lead qualification system that reads web signals and company context to produce what it calls 'decision-ready lead recommendations.' The emphasis the company places on precision over volume is notable — most lead gen tools optimize for output quantity. Leadpoet is betting that quality filtering is the actual scarce resource.

A few details from a February 2026 company fact sheet stand out. Leadpoet reached a $1 million annualized run rate within its first quarter after launching. It secured backing from DSV Fund and Astrid. And it supports private deployments — customers can run intent scoring and outreach generation entirely on their own data, without exposing it to a third-party vendor. The company also uses Bittensor, a decentralized blockchain-based AI network, for part of its model infrastructure. Ramesh said the competitive, decentralized structure can improve models faster than a centralized development roadmap. Leadpoet is also a member of NVIDIA Inception, Nvidia's startup support program.