Series · Part 4 of 6
AI + Blockchain: When Intelligent Software Meets Programmable Money
What happens when AI systems no longer only analyse data but can interact with digital financial infrastructure – within clearly defined rules.
Two technologies, one missing link
Artificial intelligence is very good at analysing information and making recommendations. Programmable money makes it possible to execute payments through software. So far these two developments have mostly run in parallel. The interesting question is what happens when they are connected.
Put simply: AI provides the intelligence, digital infrastructure provides the execution. An AI system that can recommend an action is useful. An AI system that can carry out that action – within strict limits – changes what automation can mean in finance.
Automated billing and treasury
Consider a company that receives hundreds of invoices every week. Today, software can already read invoices and match them to orders. With programmable payment rails, the next step is an AI system that checks the invoice against the contract and the delivery, flags discrepancies for a person, and schedules payment automatically when everything matches.
In treasury, AI models could forecast cash needs across subsidiaries and move funds on shared ledgers at any hour, keeping balances where they are needed. Humans set the policies – limits, approved counterparties, approval thresholds – and the system operates within them.
Smart contracts with smarter inputs
Smart contracts execute exactly what they are programmed to do. Their weakness is that they depend on reliable inputs: whether a shipment arrived, whether a service level was met, what a price was at a given moment. AI can help interpret complex data – documents, sensor readings, images – and turn it into structured inputs.
This combination must be designed carefully. If an AI model feeds a wrong input into a contract that pays automatically, the error is executed just as reliably as a correct result. Verification, multiple data sources and human checkpoints for larger amounts are essential.
Machine-to-machine payments
Many devices already communicate with each other. Programmable money would allow them to pay each other too: an electric car paying a charging station per kilowatt-hour, a sensor selling data in small increments, or cloud services settling usage by the second.
Traditional payment systems are not designed for millions of very small transactions. Some digital infrastructure is, which makes these scenarios technically conceivable. Whether they become common depends on costs, standards and whether businesses and consumers actually want them.
AI agents that initiate transactions
The most discussed idea is the AI agent: software that acts on behalf of a person or company, for example booking travel within a budget, renewing subscriptions or rebalancing payments between accounts. For such an agent, access to programmable money is what turns a suggestion into an action.
The crucial phrase is 'within clearly defined rules'. Spending limits, whitelists of allowed recipients, time windows, mandatory confirmations and complete audit trails are what make this acceptable. Without them, delegating financial actions to software would be a risk rather than a convenience.
Trust, responsibility and regulation
Connecting AI to money raises questions that technology alone cannot answer. Who is responsible if an agent makes a costly mistake? How can decisions be explained afterwards? How do existing rules on payments, consumer protection and anti-money-laundering apply to software acting autonomously?
Regulators, banks and technology companies are only beginning to work through these questions. Early applications will likely be conservative: internal business processes, small amounts, strong oversight.
Why this is worth watching
AI and programmable money are each significant on their own. Together they could shift automation from 'software that helps people decide' to 'software that carries out decisions people have defined'.
That future is not here yet, and it should arrive carefully. But understanding how the two pieces fit together helps to see where finance may be heading – and which safeguards will matter most along the way.
This article is for educational purposes only and is not financial, investment, tax or legal advice.
