Your Next Customer Doesn't Have a Pulse

For centuries, financial infrastructure has been built for one type of customer: humans. Humans with passports, tax IDs, jobs, addresses. That world is changing.

It's actually already here, just not evenly distributed. Right now, software is triggering billions in economic activity:

• Algorithmic trading drives 60–70% of equity markets and the majority of the $7.5T+ daily forex volume.

• Programmatic ad platforms execute billions of automated bids daily across Google, Meta, and other networks, with some platforms processing hundreds of thousands of requests per second.

• MEV bots, which are autonomous software competing and transacting for profit, extract hundreds of millions annually from Ethereum.

• Supply chain algorithms automatically trigger millions of reorders, with Amazon's systems alone continuously generating purchase orders every second.

In each case, the "customer" is technically software, but the payment rails, compliance checks, and risk models were all designed for humans.

The Autonomy Adoption Curve

We're moving fast from:

• Stage 0 - Basic Automation: Fixed rules execute predictable tasks such as recurring payments, monthly payroll, automated expense reimbursements. No AI needed.

• Stage 1 - Human in the Loop: AI analyzes spend data, flags unused software licenses, identifies contract overpayments, and suggests cost reductions. Finance teams review and approve each recommendation before execution.

• Stage 2 - Human on the Loop: AI acts autonomously within set parameters: auto-approving expenses under policy limits, blocking non-compliant purchases, canceling redundant subscriptions, processing standard invoices. Finance teams monitor for exceptions and can override.

• Stage 3 - Fully Autonomous: AI handles end-to-end finance operations independently: vendor negotiations, budget reallocation, cash flow optimization, financial forecasting with minimal to no human involvement.

Stage 0 is mature and commoditized. Stage 1 is where differentiation is happening right now. AI that flags unused software and suggests cost reductions is becoming the competitive edge. Stage 2 is emerging as leaders build automated enforcement and controls. The real frontier is Stage 3: fully autonomous agents. While this exists in DeFi, the complete infrastructure for regulated finance is still being built.

Why the Current Stack Breaks

Our financial rails assume:

• Identity requires a government-issued ID.

• Transactions are occasional and reviewable.

• Risk should be modeled on human behavior.

With AI customers, these assumptions collapse. KYC, AML, fraud detection, credit scoring, and settlement all have to be rebuilt from first principles.

The Opportunity

We’re already seeing AI agents influence billions in spend. The next step is letting them initiate more of that spend directly, at machine speed. The winners will be those who:

• Build instant, programmable payment rails (potentially stablecoin-based).

• Create machine-to-machine identity and trust frameworks.

• Develop risk assessment for automated transaction systems.

• Deploy compliance tech that operates at machine scale.

The next Stripe may not primarily serve humans paying humans. It will serve machines transacting with machines on behalf of humans and eventually, with increasing autonomy.

Scenario to Consider

Your company’s AI agent manages cloud spending across five providers, automatically optimizing for spot prices and purchasing $10K/month in compute. Today, no one knows who should hold its payment credentials, who is responsible if it overspends, or how it would access credit. However, because the financial and operational benefits are clear, and parts of the infrastructure already exist, controlled experiments will gradually move into production — starting with low-risk spend categories and expanding as reliability, trust, and regulatory acceptance grow.

What's Being Built Right Now

Several building blocks for AI-native financial infrastructure are already emerging, though adoption remains partial:

• AI-driven spend management platforms such as Ramp and Brex use AI to flag cost-saving opportunities and optimize recurring expenses. While transactions still require human approval, these tools demonstrate how AI can increasingly influence financial decision-making.

• Programmable payments and settlement infrastructure such as Circle's USDC stablecoin and blockchain networks like Stellar and Coinbase's Base enables fast, automated payments and forms a foundational layer for machine-initiated transactions. While adoption in traditional finance remains limited, these systems are proving out the core infrastructure needed for autonomous transactions.

• Autonomous trading infrastructure, including DeFi protocols like Uniswap and SushiSwap, as well as automated trading bots, already operates independently on-chain. These systems remain largely disconnected from regulated financial infrastructure, highlighting a gap for AI-native, compliant transaction capabilities.

• AI-powered compliance and RegTech tools such as Alloy, Unit21, and Sardine can monitor high-volume transactions and detect complex patterns. Today, they still rely on human oversight and are designed around human customers rather than autonomous agents.

What's Missing

Despite these early building blocks, infrastructure for truly autonomous finance doesn't exist:

• Agent Identity & Trust: Companies want AI to handle routine financial tasks independently: paying vendors, managing subscriptions, optimizing cloud spend. But every financial system still requires a human account holder legally responsible. Without identity frameworks for AI agents, humans must approve every transaction, defeating the purpose of automation.

• Payment Infrastructure: AI could save millions by optimizing resources in real time: switching cloud providers for better prices, scaling compute during traffic spikes, purchasing API calls only when needed. But traditional payment rails have fee structures that make micropayments uneconomical and settlement times that prevent real-time optimization. Companies need infrastructure that handles any transaction size cost-effectively with instant settlement at machine scale.

• Custody & Asset Management: Companies want AI agents to manage spending within budgets, but there's no legal framework for how AI "holds" money or who's liable for mistakes. Without clear custody solutions, legal teams block autonomous spending, keeping AI advisory only.

• Machine-Scale RegTech: A single AI agent can generate more transactions in a day than a human does in a year. Compliance systems built for human behavior would flag normal AI activity as suspicious. Builders need monitoring systems that understand machine behavior to deploy AI at scale.

The financial system is about to gain a billion new customers. They won't be human. The companies building infrastructure for them will own the next era of finance.

-Kiswana