On April 23, the world’s first private blockchain transaction executed by an AI agent settled on Canton Network, marking a major milestone in the development of p402; a Canton-native standard protocol for AI agents, developed by Cambridge University researcher, Yash Bharti, and funded through Cantor8’s official research program. The purpose of p402 is to provide regulated institutions with access to the benefits of blockchain-based AI agents in a way that maintains confidentiality on sensitive information.
In March 2026, Cantor8 announced that it had awarded Yash Bharti a grant to support cutting edge research into standardising agentic AI systems in the Canton Network ecosystem. This new announcement reflects the first results of that initiative coming to fruition.
The transaction is momentous because, until now, every meaningful attempt to establish blockchain rails for AI agents has taken place on fully public blockchains. This might be fine for retail users executing small transactions, but the uncompromising transparency is wholly unsuited for institutional participants, for whom confidentiality around sensitive transaction data is non-negotiable.
The milestone transaction was initiated and executed by a Claude agent and the asset transferred was USDCx; a 1:1 USDC-backed stablecoin, developed with Circle's xReserve, and built to bring confidential, institutional-grade dollar settlement to the Canton Network.
What is p402?
p402 can be thought of as analogous with ERC-8004 on Ethereum, insofar as it acts as a de facto ‘interaction/trust layer’ for autonomous AI agents, enabling them to identify, verify, and interact with each other on blockchain rails.
However, p402 features several key differences to the likes of ERC-8004, the most important of which is that agents operating through p402 inherit Canton Network’s unique advantages.
Especially significant here are Canton’s in-built privacy and compliance features which make it the only blockchain suited to host an institutional-grade AI agent ecosystem.
p402, much like ERC-8004, will consist of three key registries, all of which are crucial to fostering an institutional-grade ecosystem of functional AI agents on Canton:
Agent Capability Registry: Provides a DAML-based template through which institutional parties register their agents, including service capabilities, terms of service, and privacy specifications.
Service Request Registry: Designed to match agent end users with agent providers, effectively offering a privacy-preserving marketplace for the use of Canton-native agentic AI.
Service Record Registry: Offers a compliant, privacy-preserving interaction history for interactions between multiple parties, detailing outcomes, performance metrics, and more, thereby allowing agent systems to establish reputation.
Why This Matters: Solving AI Agents’ Biggest Problem
The issue being addressed by p402 was flagged this month by world-leading venture capital firm, a16z. In its paper, The missing infrastructure for AI agents: 5 ways blockchains can help, a16z wrote:
“What’s missing is a common identity layer, the equivalent of SSL for agents, that standardizes coordination across platforms. While there are prominent attempts to solve this today, those approaches are fragmented… There is still no broadly adopted, interoperable way for one agent to prove to another who it represents, what it’s allowed to do, and how it gets paid.”
This was presented by the firm as the number one issue for AI agents that blockchain rails are poised to solve, and it is a problem that p402 is tackling head-on.
Through p402’s complete registry suite, AI agents will soon have a trusted system by which they can showcase their capabilities, build trust and demonstrate track record, and clarify terms and payment agreements.
Canton’s Unique Advantages
The fact that p402 is bespoke to the Canton Network, allows it to take matters a step further.
Not only does it provide a ‘way for one agent to prove to another who it represents, what it’s allowed to do, and how it gets paid’, it also ensures that agent actions and their beneficiaries maintain privacy over their activity, while still able to share necessary data with regulators and compliance officers when required, through a complete audit trail.
This is something that is simply not possible on public blockchain rails, where many pre-existing AI agents already operate.
“The payment demo alone is hugely significant. Agent-initiated payments on a privacy-preserving public L1, under institutional custody controls, is a capability that does not exist on other chains today”. Said Philip Kaddaj, CEO & Co-founder at Cantor8.
“On transparent chains, every agent payment leaks sensitive information around counterparties, transactions, and timing. On Canton, it doesn't.”
Regulated sectors and institutions cannot afford to risk leaked positions, front-run trades, and exposed counterparties, and this innovation allows for the use of AI agents to become a meaningful option for institutional participants who require strict confidentiality on their activity.
The transaction executed on April 23 settled atomically, with selective disclosure and a full audit trail - the same privacy model that already carries trillions in tokenized Treasuries and repo financing on Canton.
Cantor8 Research & University Program
Cantor8 is the leading infrastructure provider for the Canton Network ecosystem. Through its suite of products (which covers solutions for custody, private transfer, asset issuance, and more), Cantor8 is able to effectively bridge the gap between the unique features that Canton Network offers, and the regulated/institutional participants that want to access them.
As part of its mission, Cantor8 has established partnerships with some of the world’s most respected academic institutions, including organisations from:
- Cambridge University
- Oxford University
- Imperial College London
- UC Berkeley
Cantor8 hosts regular events, lectures, and hackathons in order to expose this world-leading talent to Canton Network and its role in the future of finance. As part of its university program, Cantor8 operates a robust research funding program - an initiative which itself has made the development of p402 possible.
We are always keen to hear from academics and researchers looking to bring value to both the Canton Network ecosystem, and the blockchain industry more broadly. If you are looking for support on a research project, we want to hear from you - Please reach out to jonny@cantor8.tech.
