Wall Street's AI Agents Are Already Doing the Job

Earlier this year, Crypto.com let go of 12% of its workforce. The stated reason: AI-driven efficiency. Around the same time, Block — Jack Dorsey's payments company — announced plans to cut more than 4,000 people, roughly 40% of its staff, citing improved AI models. Crypto research firm Messari replaced its CEO and declared itself an "AI-first company." These aren't isolated restructurings. They're signals from the front line of a shift that's already underway in finance.

What Nasdaq Is Actually Doing With AI

Pranav Ramesh, head of options research at Nasdaq and co-founder of AI startup Leadpoet, has a closer view than most. Speaking to CoinDesk, he was unusually direct: "AI will replace many jobs now performed by humans," he said, pointing specifically to lower-level software development, customer service, and analyst roles as the first to go.

This isn't theoretical for him. Over the past 18 months, Nasdaq has deployed AI agents across market surveillance, compliance, and market microstructure analysis. Its subsidiary Nasdaq Verafin runs what it calls an "Agentic AI Workforce" — automating high-volume, low-value anti-money laundering compliance tasks. Back in 2023, Nasdaq launched the Dynamic M-ELO order type, the first AI-powered order type approved by the SEC, using a model with more than 140 factors to adapt to real-time market conditions.

The shift Ramesh describes isn't just about new tools — it's about trust. Earlier AI systems hallucinated too frequently for sensitive enterprise workflows. But over the past six months, he argues, that threshold has been crossed. Today, many of Nasdaq's systems operate with AI handling most of the analysis and execution, while humans remain only at the final approval stage.

Crypto Will Move Faster Than Traditional Finance

Ramesh sees crypto trading platforms as the likely leaders in deploying retail-facing AI agents — think position analysis, trade suggestions, and execution support built directly into consumer apps. The reasons are structural: lighter regulation, a user base already comfortable with new technology, and markets that never close.

Sponsored

Advertise with Us

[email protected]

Sponsored

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

He's careful to stop short of calling this full autonomy. The model he describes — and the one operating at Nasdaq — keeps humans at the final decision point. But as AI handles more of the upstream analysis, the question of what that human checkpoint actually means becomes harder to answer.

From Observation to Startup

Ramesh turned that observation into a business. Leadpoet, which he co-founded with Gavin Zaentz after the two met at Nasdaq, is an AI-powered lead qualification platform. The problem they kept running into: outbound sales tools could generate static prospect lists, but identifying real buying intent still required manual research. Leadpoet claims to automate that gap — turning web signals and company context into "decision-ready lead recommendations."

The company reached a $1 million annualized run rate in its first quarter after launch and has backing from DSV Fund and Astrid. It runs on Bittensor, a decentralized blockchain-powered AI network, and is a member of NVIDIA's Inception program for AI startups. Ramesh says the decentralized structure appeals to him because it can improve models faster than a centralized roadmap — a philosophy that mirrors how he sees AI adoption in finance more broadly.

The Jobs Question Nobody Wants to Answer Directly

Most executives soften the labor displacement story. Ramesh doesn't. He frames job losses not as a future risk but as a current observable trend. And the data supports him: Crypto.com, Block, and Messari are three major names in crypto and fintech that have already cited AI as a direct driver of significant headcount reductions — all within the past several months.

For investors, the cost structure implications are significant. A financial platform that has internalized AI agents operates at fundamentally different margins than one that hasn't. That gap is likely to widen over the next three to five years, making AI adoption less of a competitive advantage and more of a baseline requirement for survival.

For regulators, the picture is more complicated. The SEC approved Nasdaq's AI-powered order type in 2023, but that was a single, well-documented system. As AI agents proliferate across compliance, surveillance, and trading — and as crypto platforms move faster than traditional finance — the regulatory frameworks built around human decision-making may struggle to keep pace.