AI becomes the decision layer of the digital economy. Settlement, ownership, and provenance still have to run on something.
Paul Soliman is the Founder and CEO/CTO of Hacktiv Colab Inc. and Chairman and Group CEO of BayaniChain, where he leads initiatives in blockchain, enterprise tech, and digital nation-building. He also serves as CTO of Blockfy, driving innovation in decentralized finance solutions in the Philippines. More articles of Paul Soliman on BitPinas here.
The question being asked in Manila now is whether blockchain still matters when AI has taken the capital, the talent, and the attention. The more useful framing is that AI may be what finally makes the infrastructure useful. An AI that answers questions needs a model and an API key. An AI that acts is a different thing entirely, because it becomes an economic actor, and economic actors need to hold value, authorize payments, prove what they did, and operate within limits set by someone else.
Useful AI Framing
Consider what changes when an agent is given an objective instead of a prompt. Find the supplier, negotiate the price, buy the goods, monitor delivery, release payment when the conditions are met. Every clause in that instruction assumes infrastructure that does not exist for machines today. Most financial systems were designed around people, corporations, institutional accounts, and intermediaries that authenticate and authorize activity at different points in the transaction. They were not built for software making thousands of small, conditional economic decisions across networks. Blockchain systems were designed around programmable rules, cryptographic authorization, shared state, and settlement infrastructure that can operate continuously.
That difference is now being studied seriously. Researchers at the Cambridge Centre for Alternative Finance are examining autonomous software executing transactions on-chain, with stablecoins emerging as a practical medium because agents need predictable units of account rather than assets whose purchasing power can move several percentage points while a transaction is being negotiated. (Cambridge Judge Business School) Coinbase Institutional has made a related argument, that the durable position in the agent economy may not be the agents themselves but the infrastructure underneath them: wallets, stablecoins, settlement layers, and payment rails. (Coinbase)
Watch what a regulated bank actually allowed an agent to do
On 18 May 2026, Sygnum became the first regulated Swiss bank to test live blockchain-mainnet transactions using an AI agent. What it withheld matters as much as what it granted. The agent interpreted plain-language instructions, planned the transaction sequence, examined smart contracts, and identified risks, but authorization stayed with the human. Transactions were approved and signed through a self-custodial wallet, and the private keys never left client control. The agent could prepare the economic action. It could not control custody. (Sygnum)
That distinction is the whole design. Agents propose, humans authorize, rails settle. Fully autonomous machine wallets already exist experimentally, but what remains immature is their deployment inside regulated financial institutions, with meaningful balances, real liability, and governance structures that survive outside a demo environment.
The same restraint applies with more force here. The Bangko Sentral ng Pilipinas closed the regular application window for new Virtual Asset Service Provider licences in September 2022, and from 1 September 2025 it continued the moratorium, subject to reassessment as conditions evolve. The direction is clear enough. Autonomous finance in the Philippines will not arrive by ignoring regulation. It will arrive by designing machine activity around it.
The Philippine numbers make the argument better than the theory does
Moving that money is not free, and the cost is the part of the story that rarely makes the headline. The World Bank measured the global average cost of sending a remittance at 6.36 percent in the third quarter of 2025, still far above the UN target of 3 percent by 2030. (World Bank Remittance Prices Worldwide) Philippine corridors vary and many are cheaper than that average, but apply the global figure to a flow of this size and the implied cost runs past two billion dollars a year. That is money leaving the corridor before it reaches a household, extracted by the rail rather than by anyone who did the work.
Then, on 1 January 2026, a new U.S. federal excise tax took effect on certain remittance transfers. Under Section 4475 of the Internal Revenue Code, a one percent tax applies when a remittance is funded with cash, money orders, cashier’s checks, or similar physical instruments, while transfers funded directly from qualifying bank accounts or with debit and credit cards fall outside it. Proposed Treasury regulations draw the same line between physical and non-physical funding methods. (IRS)
That is an infrastructure story rather than a tax story. The excise sits on top of whatever the corridor already charges, and what triggers it is not the amount sent or the country it goes to, but how the sender funded the transfer. Given that nearly two in five dollars of Philippine remittances originate in the United States, a policy written in Washington has placed a direct price on one funding rail and not another. Millions of Filipino households will feel the difference without ever learning the regulatory reason behind it. The rail has become a household expense.
The country is less unprepared for this than people assume. A peso-backed stablecoin has already completed a Bangko Sentral sandbox pilot and begun exiting the framework in mid-2025, backed one-to-one by cash and cash equivalents according to issuer disclosures, with the sandbox process covering BSP reporting, holder disclosures, and independent audits including proof-of-reserve verification. On 5 July 2025, SEC Memorandum Circulars 04 and 05, Series of 2025 took effect, creating a more explicit framework for crypto-asset service providers. A CASP must now be established domestically, maintain a physical presence in the Philippines, and hold at least ₱100 million in paid-up capital, excluding crypto-assets. (SEC)
None of this means the Philippines suddenly discovered crypto regulation in 2025, since the BSP had established a VASP licensing regime years earlier. What changed is that the perimeter became clearer, the capital requirements became harder, and the distinction between compliant infrastructure and speculative activity became more difficult to ignore.
Government has started moving at a pace that is hard to ignore
Philippine public-finance blockchain work began before the national budget went on-chain. In July 2025, the Department of Budget and Management launched a blockchain verification system for budget-release documents including Special Allotment Release Orders and Notices of Cash Allocation. The legislature followed. Speaker Faustino “Bojie” Dy III announced that the House of Representatives would move to a fully paperless system and institutionalize blockchain beginning 2026, in partnership with DICT, and stated that the move makes the Philippine Congress the first legislative body in Asia to adopt blockchain for governance. (BitPinas) The ₱6.793 trillion 2026 General Appropriations Act was signed on 5 January 2026 and recorded under the Digital Bayanihan Chain, covering the GAA of the whole country.
DICT has separately claimed a stronger milestone, that the Philippines is the first country in the world with a fully on-chain national budget. Local analysts have questioned parts of that framing, including the security claims made around it. The first-in-Asia legislative claim is the sturdier one, and it is the one that matters for what follows.
Disclosure: BYC Ventures built the blockchain implementation for DICT.
What the project demonstrates has little to do with digital assets. When agents start consuming public records at scale, and they will, the important question stops being whether a document is online and becomes whether the machine can prove that the document it read is the same document the government originally published. Availability is not provenance. A PDF can be hosted publicly, a database can expose an API, and a government portal can publish thousands of documents without any of that establishing what the record said on the day it was issued.
When those records start informing procurement systems, credit models, compliance engines, and AI agents, someone eventually has to answer whether the underlying record was altered after publication. Cryptographic signatures, tamper-evident logs, and append-only systems are among the practical ways to answer that question. Blockchain becomes particularly useful when several institutions need to independently verify the same record without relying on one database owner to certify its history. That is a very different proposition from trading tokens.
Another important signal is Senator Bam Aquino’s CADENA Act, which became one of the clearest examples of government experimenting seriously with blockchain. The measure proposes using a tamper-evident digital ledger and public budget portal to make government spending more transparent and auditable. More than the technology itself, CADENA showed that blockchain was beginning to enter mainstream policy discussions not as speculation, but as infrastructure for accountability.
Most tokens..
Being long the infrastructure is not the same thing as being long the assets, and the Philippines learned that locally and expensively. Play-to-earn spread across Philippine communities in 2021, and for a period gaming tokens became meaningful sources of household income. Then prices collapsed. People who thought they were holding productive digital assets discovered that much of the value depended on a constant flow of new participants, speculative demand, and token incentives that could not survive indefinitely. A lot of Filipinos were left holding tokens whose value had collapsed and income streams that had disappeared.
The same structure will return wearing different vocabulary. Over the next two years, a great deal of what gets marketed as AI plus crypto will exist because those three words raise capital. That does not make the underlying infrastructure thesis wrong. It makes the distinction more important. Payments, settlement, identity, ownership, provenance, machine-to-machine commerce: projects that can explain in one sentence what they do better than the existing alternative have a chance of surviving the cycle. The rest will not, and should not.
The endpoint is invisibility
Nobody wants to execute an on-chain stablecoin transfer. They want to send padala. A business owner will not say she is tokenizing an asset, she will say she is raising capital. A procurement officer will not say she is verifying a Merkle proof, she will say she knows the document is authentic. An AI agent will not care whether the rail underneath it is called crypto, fintech, Web3, or something not yet named. It will choose whatever system can move value, enforce constraints, settle the transaction, and leave behind evidence that can be verified later.
That is the version worth building in the Philippines. Not only Filipinos trading tokens, but households moving money on rails that cost less, businesses settling faster, government records that machines can verify, and agents able to participate in the economy without every action having to be checked by hand. AI may become the decision layer of the digital economy. Decisions still need rails, and the Philippines has already started building them.
This article is published on BitPinas: AI Will Act. Blockchain Will Settle. The Philippines Is Preparing for Both.
What else is happening in Crypto Philippines and beyond?
