NEO World, NEO Finance: Could AI Agents & Programmable tokens be used in banks, and why leading banks are already piloting?

Thinking note for Belgrade workshop next week.

I don’t like to talk to many people about new concepts, but when I know that there is an opportunity to give value to the people when they see a clear path of how the future will unfold and how they can, with existing capabilities, make their organization more profitable and resilient, it is hard to resist.

I am excited about this workshop for more reasons.

The venue is perfect, a private club better than any I’ve been a member of so far.

From all the countries I have lived and worked in, and all the clubs I was a member of or just a guest.

This is our new Belgrade base.

So I can promise my guest that the venue will be perfect.

The Future of Banking in the NEO World

In the workshop, we will explore current and future outlooks, as well as possible roadmaps, for banks, businesses, and people in the emerging machine-driven economy.

If you work in banking, you probably see the signs of the incoming change.

For decades, our world and your world were built for a human pace.

A relationship manager met a client.

A company sent an invoice.

A payment batch went out at the end of the day.

Everything important happened between 9 and 17h.

NEO World

That world isn’t collapsing overnight – but it is changing fast.

We are entering what we call the NEO World.

New Economic Operating System where:

AI agents will negotiate contracts in milliseconds.

Money will move 24/7 across cloud and edge networks.

Machines will pay machines for services with almost no human involvement.

This is not science fiction.

Corporate treasuries, leading world banks, cloud providers, and IoT networks are already piloting and learning to understand what the future will unfold.

So the question for banks is no longer: “Will this affect us?”

but:

“What role do we want to play – the leader or follower.

The Two-Layer Future of NEO Finance can be illustrated with a simple picture.

The Macro Layer – The Human World

This is where banks are strong today:

Salaries, mortgages, trade finance SEPA, SWIFT, RTGS, card schemes, hundreds, thousands, millions of euros

It’s a world of documents, signatures, quarterly reports, and board approvals.

It is important.

It’s not going away.

The Micro Layer – The Machine World

This is the NEO frontier.

Imagine billions of micro transactions every day, between €0.001 and €1.00:

An AI agent paying for a single API call, an IoT device buying 5 seconds of connectivity.

A model paying per millisecond of GPU time.

Economically, this is huge.

But on today’s rails, it’s almost invisible:

Too small, too frequent, too complex to reconcile profitably.

So what happens? We bundle it into big fees, write it off, or never start at all.

That’s what we call the micro-transaction barrier.

If we want growth in NEO World, this is the wall we have to break down.

Why Stablecoins, programmable settlement tokens & Tokenized deposits matter?

When machines start transacting, they need money that behaves more like them:

Always on, Programmable, and predictably valued.

This is where stablecoins, programmable settlement tokens, and tokenized deposits come into play.

Think of them as a bridge between two worlds: For banks, this is not “the enemy”.

It’s a chance to take what you already do best – trust, compliance, custody – and package it in a form that AI agents and machines can actually use.

A Glimpse into the Machine Economy

In Project 4028ra, within the HashNET, we’re prototyping this “machine layer” in practice. Now imagine this: one “normal” corporate payment triggers a whole hidden economy of tiny, instant machine-to-machine payments between AI agents, devices, and services – all running in the background, without a single invoice or email reminder.

This is the kind of scenario we believe we’ll see in the next wave of banking innovation.

In Belgrade, I’ll walk through how it could look in practice.

If there’s interest in going deeper, there are hard questions we need to explore.

This vision is exciting, but it’s not a fairy tale.

This is not about replacing banks.

It’s about upgrading them to be the sovereign, trusted backbone of the machine economy.

This is not about replacing people.

But to make people who know more and want more to be more resilient. Remember the hard truth: Fast and smart eats slow.

HERE IS WHAT AI THINKS ABOUT THIS: 🙂

Your vision of a “NEO World” with a new economic operating system is not just a theoretical concept; it is a strategic reality that leading financial institutions are actively building today. The convergence of AI Agents and programmable tokens represents the most significant opportunity for banks to future-proof their operations, unlock new profit pools, and become the sovereign backbone of the emerging machine-driven economy.

The table below summarizes how these two technologies combine to target core challenges and opportunities in modern banking.

TechnologyCore Disruption/ApplicationPrimary Driver for BanksAI Agents

Performing tasks and solving issues autonomously; shifting from reactive helper to proactive financial agent.

Defensive & Offensive: Combatting revenue loss from disintermediation and capturing new profit potential exceeding $370 billion annually by 2030.

Programmable Tokens(Stablecoins, Tokenized Deposits): Digital value that executes automatically under set conditions, with compliance and control built in.

Strategic Necessity: Enabling 24/7, conditional, and transparent transactions that are essential for the machine economy, ensuring banks remain the trusted node.

The Strategic Drivers: Why Banks Can’t Wait

Banks are piloting these technologies not out of fear, but because of a clear and compelling economic imperative. Several critical factors drive the transition:

• Breaking Customer Inertia, Unleashing Optimization: Traditional banking revenue, particularly from deposits and credit cards, relies heavily on customer inertia. AI agents fundamentally break this model. They will not be loyal to a brand but to an outcome, automatically optimizing for the best rates and rewards. This could compress net interest margins by 30-50 basis points as agents sweep idle cash to high-yield accounts and erode credit card revenue by strategically routing payments to avoid fees and maximize rewards.
• The Rise of the Machine Economy and the Micro-Transaction Barrier: Your concept of a “Micro Layer” is astute. The future will involve billions of machine-to-machine micro-transactions that are uneconomical on today’s rails due to their small size, frequency, and complexity. Programmable money is the key to unlocking this frontier, providing the “always-on,” granular, and logically-driven payment layer that machines require to transact efficiently.
• The $370 Billion Profit Opportunity: Inaction is not a viable strategy. BCG identifies over $370 billion in annual profit potential from AI by 2030. Leaders who successfully embed these technologies will alter the industry’s competitive dynamics, leaving slower movers increasingly irrelevant. S&P Global emphasizes that the next 3-5 years will be determinative, creating a lasting “AI gap” between leaders and laggards.

The Winning Combination: Use Cases in Action

The true power is unleashed when AI agents are empowered with programmable tokens. This creates a seamless, autonomous financial operating system.

• For Corporate Treasuries: Imagine an AI agent that manages corporate liquidity in real-time. Using programmable tokens, it can automatically sweep excess cash across global subsidiaries into higher-yielding tokenized money market funds, execute cross-border supplier payments the instant goods clear customs, and use tokenized assets as collateral for instant loans—all without human intervention. This transforms treasury from a reporting center into a live control tower for global liquidity.
• For Autonomous Commerce: An AI travel agent can be instructed: “Book a trip to Paris for two, under $2,000, and pay with the best card.” The agent researches options, makes bookings, and at the point of payment, it doesn’t just recommend a card—it uses a programmable payment token to dynamically route the payment to maximize rewards or even initiate a direct account-to-account payment to bypass card interchange fees altogether.
• For New Markets and Services: Banks can use stablecoins to enable U.S. corporations and their foreign subsidiaries to hold USD offshore, providing a hedge against local currency volatility and creating new deposit streams. Similarly, they can power merchant loyalty programs in which the stored value is a regulated, bank-issued token rather than a risky merchant-specific balance.

The Proof: Leaders Are Already on the Move

This is not theoretical. The world’s most prominent financial institutions are already in advanced stages of piloting and deployment, validating the path forward.

• JPMorgan has launched JPM Coin on a Public Blockchain, a tokenized deposit that combines the programmability of blockchain with the trust of a regulated bank for instant, conditional B2B payments.
• Citi is exploring programmable money for real-time reconciliation and embedding compliance directly into transactions, shifting oversight from a post-event process to a live, always-on safeguard.
• A Global Consortium including DBS Bank, Standard Chartered, and J.P. Morgan is backing the Partior network to create interoperability between different banks’ digital currencies for cross-border settlements.
• European Banks are leading in deployment, driven by supportive open banking regulations (PISP/AISP licensing) that provide the ideal groundwork for agentic systems to operate.

The Roadmap: From Pilot to Sovereign Backbone

Becoming an AI-first, programmable bank does not require a reckless “big bang” migration. A phased, pragmatic approach de-risks the journey and builds momentum.

• Phase 1: Foundation and Pilots (0-12 months): Begin by targeting internal inefficiencies or low-risk, high-value customer segments. Use cases include autonomous bill payment, smart expense management, or internal tokenized cash transfers to accelerate settlement between subsidiaries. The goal is to build foundational capabilities and demonstrate quick wins.
• Phase 2: Scaling and Integration (12-24 months): Expand to more complex applications like supply chain automation and AI-driven cash sweep programs. Focus on integrating systems to break down data silos and begin offering programmable treasury services to your most forward-thinking corporate clients.
• Phase 3: Invention and Leadership (24+ months): This is where you reinvent your role. Evolve from a financial intermediary to a provider of “Trusted Programmability”. Offer your balance sheet and regulatory compliance as a secure platform upon which your clients—and their AI agents—can build their own sophisticated financial applications, securely embedded within the ecosystems where they live and work.

Your perspective on how to make organizations more profitable and resilient by embracing this change is precisely the conversation leaders in your workshop should be having. The technology is ready, the economic incentive is clear, and the first movers are already defining the next era of finance.

I hope this provides a solid foundation for your workshop in Belgrade. If you have more specific ideas to refine, I am here to help you explore them further.

In the end, and this is from Dražen Kapusta: You should not trust AI 100% but use it as a helpful tool. At least currently. Who knows what the future will unfold?