Order the book

Autonomous Economy · Aashiesh Chopra

Search for the book on your preferred store. Direct ordering links are pending.

Search Amazon

Insights & Research

How banking changes when decisions, money, and controls move together.

Research for banks and credit unions on agentic AI, payments and digital assets, and how governance, cybersecurity, and risk evolve as financial decisions and money move faster.

How banking is evolving

Four economic operating models explain how the unit of control shifts from a standardized process to an outcome coordinated across systems.

  1. Industrial Economy

    Unit of control · Process

    Organizations scale work through labor, machinery, and standardized processes.

    Banking example A branch team reviews and approves loans through a standard manual process.

  2. Digital Economy

    Unit of control · Transaction

    Software digitizes interfaces, records, and individual transactions.

    Banking example A customer opens an app and schedules a transfer. The bank executes the transaction.

  3. Agentic Economy

    Unit of control · Task or goal

    An agent interprets a goal and carries out a bounded task.

    Banking example A business asks an agent to find liquidity options for Friday payroll. It compares the options and prepares a proposed transfer.

  4. Autonomous Economy

    Unit of control · Outcome

    Authorized agents coordinate across systems to reach and maintain an outcome within defined limits.

    Banking example An authorized workflow monitors payroll funding, coordinates permitted transfers, verifies settlement, and escalates new borrowing.

How intelligence becomes economically consequential

An agent can propose an action. For a bank to act on it, someone must delegate authority, the system must know what funds are available, and the institution must be able to verify and account for the result.

See all 10 steps
  1. Intelligence

    Interprets information and identifies possibilities.

  2. Agency

    Makes it possible to pursue a goal.

  3. Delegation

    Assigns a bounded task on someone’s behalf.

  4. Authority

    Defines what the actor is allowed to do and when it must escalate.

  5. Trust

    Lets participants verify the actor and its authority.

  6. Financial state

    Makes available balances, obligations, and limits visible.

  7. Programmable value

    Lets the system move funds under defined conditions.

  8. Settlement

    Confirms that the transfer is final.

  9. Evidence

    Records who authorized the action and what happened.

  10. Autonomy

    Coordinates actions over time to reach and maintain the outcome.

Read the full Payroll Friday scenario

Article library

Browse by topic

11 articles

Bank Economics & Operating Models

AI FinOps Is Becoming Workforce Economics. Are Banks Ready?

· Perspective · LinkedIn

How should a bank authorize, control and measure the economic value of AI-agent work? The article gives CFOs, CIOs and CROs a shared framework for the decisions before, during and after execution.

Read on LinkedIn

Agentic Customers & Commerce

When Every Transaction Becomes a Negotiation

· Perspective · LinkedIn

Payments are one step in a larger decision. The article explains how an agent could compare financing, settlement and other terms within a mandate before money moves.

Read on LinkedIn

Authority, Trust & Accountability

The Board’s New Responsibility: Its Decision History

· Perspective · LinkedIn

An AI authority map gives leaders a way to ask what machines may decide, under whose authority, and who remains accountable.

Read on LinkedIn

Authority, Trust & Accountability

Human-in-the-Loop Is Not a Strategy

· Perspective · LinkedIn

Uses a hypothetical treasury payment to distinguish receiving an alert from being able to stop the action. Introduces a tiered model for delegating authority.

Read on LinkedIn

Agentic Customers & Commerce

One Customer. Ten AI Agents. Who Owns the Banking Relationship?

· Perspective · LinkedIn

When customers delegate financial decisions to several agents, banks may have to earn permission to act one decision at a time. The article introduces Mandate Share.

Read on LinkedIn

Agentic Customers & Commerce

Banking’s ChatGPT Moment: When Customers Stop Searching and Start Asking

· Perspective · LinkedIn

Shows how questions about buying a home or managing cash flow could move the customer journey from product search to financial intent.

Read on LinkedIn

Bank Economics & Operating Models

Replacing the Core Will Not Create a Real-Time Bank

· Perspective · LinkedIn

Separates core replacement from the system of action needed for real-time banking. A strong read for CEOs and CIOs evaluating modernization investments.

Read on LinkedIn

Authority, Trust & Accountability

Your AI Denied the Mortgage. Can Your Bank Prove Why?

· Perspective · LinkedIn

Uses an AI-assisted lending decision to show why institutions need to reconstruct how a decision was made.

Read on LinkedIn

Research Horizons

Quantum Computing: Banking’s Greatest Threat or Greatest Opportunity?

· Research · LinkedIn

Explores what quantum computing could mean for banking, from preparing cryptography for a post-quantum future to researching possible applications in optimization and liquidity.

Read on LinkedIn

Bank Economics & Operating Models

The Data Divide: Why Banking Leaders Must Fix Data Before Scaling AI

· Perspective · LinkedIn

Offers a 90-day plan to prioritize data, govern one high-value domain and test it in delivery.

Read on LinkedIn

Money, Liquidity & Settlement

Why End-of-Day Banking Is Becoming Obsolete and Why Continuous Banking Will Power the Age of Agentic Commerce

· Perspective · LinkedIn

Connects continuous liquidity, settlement and control to the economics of agentic commerce.

Read on LinkedIn

Research Signals

What changed, why it matters for financial institutions, and what the evidence leaves open.

Follow the decisions in Payroll Friday

Instant payments move into higher-value use cases

What changed · Fact
Federal Reserve Financial Services announced a tenfold increase in the FedNow network limit to $10 million, effective November 2025.
Why it matters · Interpretation
Larger payments can fit on instant rails, making after-hours liquidity and exception handling more consequential.
What remains uncertain
How quickly institutions will raise their own limits and adapt their controls for larger payments.
Federal Reserve Financial Services ·

Merchants gain a way to recognize trusted agents

What changed · Fact
Visa released Trusted Agent Protocol to help merchants recognize AI agents interacting with their sites.
Why it matters · Interpretation
Recognizing the agent is one part of deciding whether it has permission to act for a specific customer.
What remains uncertain
How broadly merchants will adopt it and how participants will resolve disputes about what the customer authorized.
Visa ·

Put the research to work

Use these questions to prepare a board discussion or assess where your institution can act.

Board briefings

Discuss what board members need to know about delegated authority, controls, and accountability.

Executive strategy session

Identify opportunities, risks, and decisions your institution should prioritize.