AI | Blockchain | Energy | Health | Marketing

Wall Street Bets $81B on Agentic AI: Genius or Gamble?

Wall Street has a new fix: agentic AI. These autonomous systems decide, act, and adjust without constant guidance. Market researchers think spending on them in financial services will jump from 2.1 billion dollars in 2024 to 80.9 billion by 2034, a 43.8 percent annual rise. Big numbers aside, the real story mixes breakthroughs, rules, and risks that would rattle any risk officer.

When AI turns into a colleague

Goldman Sachs has already given its GS AI assistant to about 10,000 staffers and wants every knowledge worker to have it this year. Chief information officer Marco Argenti says talking to the assistant feels like talking to another Goldman employee. The next step, he adds, is letting it finish tasks on an employee’s behalf.

Goldman is not alone. JPMorgan Chase runs AI tools for more than 200,000 people. The bank says these tools have saved roughly 1.5 billion dollars through fraud cuts, better trading, and faster credit calls. A tool named Coach AI helps advisers find answers 95 percent faster and could let them grow client lists by half within five years.

Why agentic AI is different

Standard AI answers prompts and follows set rules. Agentic AI senses its setting, makes choices, and carries out multi-step jobs with little help. Think of the gap between a calculator and a junior analyst. The World Economic Forum says these systems may change everything from trading to compliance. Gartner calls them the top tech trend for 2025 and expects them to handle at least 15 percent of routine workplace decisions by 2028.

Proof is showing up. In January 2025 Citi rolled out virtual wealth-management assistants that act on their own. IBM launched a tool that spots fraud and reacts to rule changes in real time.

The 35 million-euro reality check

Europe’s AI Act started to bite on 2 February 2025. Banks that slip can face fines of 35 million euros or 7 percent of global sales. Credit scoring, insurance pricing, and many other core bank tasks fall under the Act’s high-risk label. Even US banks are paying attention, since the EU rule book may become the global yardstick.

The hallucination trap

These systems sometimes invent facts. Top models still produce false claims at least 0.7 percent of the time, and rates above 25 percent are common. In finance, one study found hallucinations in up to 41 percent of queries. A Deloitte survey showed 38 percent of executives made wrong calls in 2024 because an AI tool fed them bad data. A single wrong split ratio in a 10-K can ripple through investment models.

Building controls in from day one

The safest rollouts bake compliance into the code. JPMorgan has about 450 potential AI uses in sight and expects that to double soon, but each use sits inside strict guardrails and human checks. Asset-and-Wealth-Management chief Mary Erdoes says the goal is to hand AI power to more staff without losing control.

Market pressure

No bank wants to lag. From 2024 to 2028 financial firms will drive one-fifth of the world’s extra AI spending. North America owned 38.4 percent of the market in 2024, yet growth is spreading fast. Researchers put the broader enterprise agentic AI market at 2.59 billion dollars in 2024 and see 46.2 percent yearly growth through 2030.

A steady rollout ahead

Deloitte expects one in four companies using generative AI to test agentic pilots this year and half to do so by 2027. Firms will start with well-defined tasks, then move to thornier ones. Supply-chain teams already use agentic AI to match sales history with weather data; finance will follow the same pattern.

Humans still steer the ship

The winning setups keep a human above the loop. AI handles the grunt work, but people keep final say on high-stakes calls. Pawel Gmyrek from the International Labour Organization puts it plainly: AI should enhance judgment, not replace it.

Agentic AI is reshaping finance. The sector can reach the promised 81 billion-dollar payoff, but only if leaders pair bold moves with tight controls. Trust has always been a bank’s currency. That will not change.

Further Reading: