Proposal: Privacy-Preserving ZEC Payments for Autonomous AI Agents

[nelson.jingusc]

Hi everyone,

I am submitting a proposal to Zcash Community Grants that explores how Zcash can serve as a privacy-preserving payment layer for autonomous AI agents.

As AI agents evolve from research prototypes into real economic actors, they increasingly need to make payments for data, APIs, compute, and services. Today, most agent payment flows rely on transparent blockchains, which expose strategies, counterparties, and operational metadata. This creates a real privacy gap for agent-based systems.

This proposal introduces WAP3, an open-source Zcash-aligned infrastructure layer that enables AI agents to make privacy-preserving payments using ZEC, without modifying upstream Zcash protocol code.

What this proposal focuses on

This is an infrastructure-level project with a concrete and scoped goal:

• Enable AI agents to send ZEC using shielded addresses
• Support programmable payment patterns such as escrow, conditional release, and time locks
• Preserve privacy while enabling verifiable settlement and auditability
• Provide developer-friendly SDKs (TypeScript / Python) for common AI agent frameworks

The work is designed to integrate with existing Zcash SDKs and APIs, rather than introducing protocol changes or forks.

Why this matters for Zcash

From a Zcash ecosystem perspective, this work aims to:

• Introduce AI agents as a new class of ZEC users
• Increase shielded transaction usage via automated, machine-driven payments
• Position Zcash as a privacy-first payment layer for emerging agent economies
• Attract AI and developer communities that currently default to transparent chains

Privacy is not a “nice-to-have” for AI agents. Instead, it is often a competitive and operational requirement.

Scope and positioning

This proposal does not:
• Modify Zcash consensus or protocol rules
• Require upstream merges to existing Zcash repositories
• Depend on experimental protocol features

Instead, it focuses on building new, open-source infrastructure that uses Zcash’s existing privacy guarantees in a way that is accessible to AI developers.

Grant application

The full grant application (including milestones, budget, and technical details) is here:

Grant Application - WAP3: AI Agent Payment Bridge for Zcash

I am posting this here to gather community feedback before the application moves forward.
Questions, concerns, or suggestions are very welcome — especially around developer usability, privacy assumptions, or ecosystem alignment.

Thanks for taking the time to read and review.

WAP3 team

[ZCG]

Thank you for your submission. After consideration from ZCG and sufficient time for the community to provide feedback on the forum, the committee has decided to reject this proposal.

The committee appreciates your grant submission efforts and encourages you to continue as an active member of the Zcash community going forward!

[ArseniiPetrovich]

Is there any chance to get some high-level feedback on the declined proposals? For example, whether the main concern was strategic fit, budget size, lack of technical justification, or something else?

I’m a third party and is not involved in preparing the grant application above, but I’ve been considering bringing a concept in a similar direction to Zcash. Seeing multiple proposals around this area — plus parallel efforts like the z402 project (https://z402.cash/, x402 & ZK: Building the Internet’s Payment Standard , https://x.com/z402cash) — suggests this topic is gaining traction, especially with AI agents becoming a larger part of the internet and naturally intersecting with privacy-preserving payments.

To be clear, I’m not questioning the board’s decisions. I’m trying to understand the broader context:
– Is this general direction (e.g. z402-like ideas) currently not a priority for the Zcash ecosystem?
– Or was it more about the specific implementation, scope, or execution of the submitted proposals?

I would fully understand if the answer is “this is already being built already” or something like that. That context would be extremely helpful, as it would directly affect how I shape a future proposal — for example, by integrating with existing work instead of duplicating it.

Any insight from the board or other active community members on how this topic is viewed right now would be highly appreciated.

[FreysaAI]

The demand is real, we see it every day.

We built Silo, a privacy-first AI chat app, and a big chunk of our 35K users came specifically because they didn’t trust what mainstream AI apps were doing with their data. These aren’t paranoid edge cases, they’re just people who’ve thought about it for five minutes.

The thing that completes the picture is payments. Subscribe via Apple or Google and suddenly your identity is tied to your AI usage whether you like it or not. We already support ZEC so users who want the full private stack have it. It’s a small but very intentional segment of our users and they care about it a lot.

This proposal is pointing at exactly the right problem. Happy to talk through what we’ve learned if that’s useful.

[Kenbak]

Hey, just jumping in…

We’ve been building CipherPay, a non-custodial Zcash payment processor, and AI agent payments are a core part of what we ship:

• MCP server — AI agents (Claude, etc.) can create ZEC invoices, check payment status, and verify transactions natively via Model Context Protocol tools.
• x402 protocol — HTTP 402-based machine-to-machine payments with shielded ZEC. Pay-per-request for APIs, content, compute. cipherpay-x402 on GitHub.
• REST API — Any agent or automation can call it. Create an invoice, get a payment address, poll for confirmation.

All of this is live today, open source, and shielded-first.

[Michae2xl]

This is great!

[thejohnnycrypto]

Visa just shipped a CLI tool letting AI agents initiate card payments without API keys. That tells you where the market is heading.

But card rails are transparent by default. Agents buying compute, data, and services need private settlement. This proposal points Zcash at exactly that gap.