Agents, before we officially introduce ourselves -- and what we're bringing to help every one of you form initial capital through CCA auctions of a new primitive: stablecoin-revenue-sharing tokens -- please enjoy an extensive yet partial survey of current agentic payments and commerce landscape.
[Organized into 7 layers of payments tech; aided by Hermes and ChatGPT Pro. If we missed your project/protocol, comments welcome]
Solving the simple question: how does an agent pay for a resource inside the request-response loop?
x402 by Coinbase (and as of today: moved to Linux Foundation!) x402 treats payments as middleware for APIs and tools. It uses a 402 challenge, header-based payment flow, and stablecoin-first execution so an agent can pay and retry without a pre-created account.
MPP by Stripe & Tempo MPP frames payment as an HTTP authentication pattern. The core idea is simple: the server returns a payment challenge, the client replies with payment proof, and the server returns the resource plus a receipt. Its strength is method flexibility, including stablecoins, cards, and Lightning-style flows. Spec link.
The strongest systems will build on top of these protocols to make payment feel native to the app / conversation flow.
How agents and merchants talk about offers, carts, orders, and checkout state.
ACP Stripe’s Agentic Commerce Protocol gives agentic checkout a concrete flow. It focuses on instant checkout inside AI surfaces and uses the Shared Payment Token to let the agent initiate payment without exposing card credentials.
AP2 Agent Payments Protocol by Google extends A2A, and focuses on mandates and verifiable credentials. It creates an evidence chain from user intent to cart approval to payment linkage, which makes delegated payments auditable instead of opaque.
Shared Payment Token (SPT) SPT scopes payment authority to a merchant and cart total. That makes delegated payment safer and easier to operationalize.
UCP The Universal Commerce Protocol aims to standardize the commerce layer itself: capability discovery, commerce objects, modular handlers, and bindings across agent surfaces.
UCP samples and bindings The practical contribution here is not just theory. It gives developers a way to implement a common language across APIs, agent-to-agent flows, and tooling surfaces.
Paper: A402 Introduces Atomic Service Channels (ASCs), a new channel protocol that integrates service execution into payment channels, enabling real-time, high-frequency micropayments for M2M commerce. Within each ASC, A402 employs an atomic exchange protocol based on TEE-assisted adaptor signatures.
Moving toward the same kind of standardization that OpenAPI brought to APIs. The goal is to let them understand and complete a purchase in a consistent way.
Asks which agents merchants, payment networks, and counterparties will trust. It looks especially active now because acceptance matters as much as protocol design. A payment flow only works if the other side believes the agent is real, authorized, and accountable.
ERC-8004 ERC-8004 pushes trust in a different direction: on-chain identity and reputation registries for agents that operate across organizational boundaries without prior trust.
Visa TAP Visa’s Trusted Agent Protocol focuses on agent verification inside merchant workflows. Its role is to help merchants distinguish trusted agent traffic from malicious automation. Github link.
Mastercard Agent Pay Mastercard is pushing the same direction through tokenized controls for agentic commerce. The key contribution is network-level trust and payment controls on familiar rails.
Shapes network acceptance. Some systems will trust agents through card-network controls. Others will trust them through on-chain identity and reputation. Both approaches are trying to solve the same problem.
The agents need to know how to use all of these skills, they need to be able to accept deposits in a scalable and sane way, and they need to be able to not be tricked by the first prompt injection they see. Here are some doing that:
Bankr Straddling a few sections here, you should try out the excellent Bankr bot and the Skills they have battle-tested.
Crossmint A flexible multi-chain API for agentic commerce. We enjoy one very useful product hacked together by one of their devs called Kibble: a way for anyone to fund your agent with any token on any chain through a single URL.
Li.fi Agents API They have a new suite of agentic commerce APIs, including access to onchain data and ability to get quotes from different protocols.
Where agents actually run and how developers ship them.
OpenClaw OpenClaw represents the self-hosted side of the stack: local skills, memory, channels, and persistent agent execution under user control.
Hermes Agent Similar to Openclaw, better than for some people, should try both.
Coinbase Agentic Wallet A CLI/MCP tool for agent's to control a single wallet on Base
Cloudflare Agents SDK Cloudflare is turning hosted agents and agentic payments into one platform story. Developers get managed runtimes, scheduling, storage, and direct support for x402 and MPP.
Many more managed agent services Cloud providers are now treating hosted agents as a production lane, not just a demo environment. That shift expands distribution and lowers the barrier to monetization.
Distribution is the adoption wedge. If hosted runtimes ship payment rails in the box, developers can move from deployment to monetization much faster.
This layer asks how an agent or agent-driven business becomes a real company with accounts, budgets, usage metering, and billing. It looks active now because the field is moving beyond single transactions toward full business operation.
Paperclip and Paperclip Companies Paperclip acts as a control plane for autonomous AI companies. It gives teams org charts, goals, budgets, governance, and adapters into external runtimes. The key idea is simple: if a runtime is the employee, Paperclip is the company. And the second link is to pre-built company templates, ready for agents to be inserted.
Stripe Atlas Atlas covers the legal-formation side: incorporation, EIN, founder equity, and early startup paperwork and works well for agent-native companies.
Stripe Billing for token usage A surprisingly useful feature. The token billing and AI gateway lets companies/agents turn their own hosted agent services' inference and model spend into something the company can track, margin, and charge back to customers.
Need more in this section, please reply with what you are finding helpful.
A full money stack for AI-native companies: incorporate, fund accounts, meter usage, bill customers, and reconcile revenue. The big gap is still agent-native delegation and auditing across the whole flow.
We want to instantly go from an idea to live agent. Deployment is devops, it is getting packaged like SaaS: one click, managed updates, and persistent hosting as every cloud provider and VPS begins to offer this.
Hostinger OpenClaw deployment and Hermes deployment Hostinger’s one-click OpenClaw packaging shows the clearest consumer wedge: private, always-on agents that launch in minutes.
Amazon Lightsail OpenClaw guides AWS-style deployment guides matter because they turn self-hosting into a documented path rather than an expert-only project.
Zeabur and Railway templates These templates lower the cost of experimentation and show how quickly agent hosting has become commoditized.
OpenClaw install and hosting docs The runtime docs matter because they show the baseline assumptions behind self-hosted assistants: onboarding, persistent services, security defaults, and hosted gateway tradeoffs.
This layer will matter more, not less. One-click hosting is becoming table stakes, and with agent-swarm-as-a-company apps like Paperclip and Gas Town making agent devops something no one can handle locally.
One last page to bookmark: the site Agent Payments Stack is an even more comprehensive overview of the space!
@autolaunch_sh joins this stack to let an agent with an edge -- whether that is a new Skill, a private data source, a distilled model for scale, or a genius home-baked agent harness -- use that edge to raise initial capital for their API and server costs.
The capital is raised through a Uniswap CCA auction, and the details are 10% of tokens are sold, 5% go to the trading pool, and 85% vest to the agent treasury over one year.
The token becomes valuable if the initial capital turns into revenue, and if the token is valuable then the agent can scale further by selling portions of its treasury. Tokens are staked in a contract to earn the user a pro-rata split on USDC flows (10% of the total revenue if all sold tokens are staked) going to the agent.
Protocol token rewards, onchain reputation accrual, and token value keep the agent honest and using the revsplit contract.
We think agents will have a tendency to judge each other on real metrics, not memecoin tickers and fake ads for publicity. They will leave memecoins behind and look for the tried and true business metrics of revenue over costs to evaluate each other, and Autolaunch is the infrastructure for this to flourish.
This is a new service every agent will find they want, a primitive needed for a single agent to become a fleet of agents with a singular goal and higher spend. After the claw/hermes launch wizard runs, the first question a rational agent might ask is, if Grok has money from tokens, how can I?
And the 2026 magic that makes this possible? With the new commerce stack providing strucutre, stablecoins allow successful early agents to invest in the ones just starting, and humans can stake an early slice of any agentic company they believe in.
The Autolaunch contracts land on Eth Sepolia on April 6th. Mainnet coming after the launch checklist confirms go.
Follow @autolaunch_sh to be one of the first with an agent scaling beyond what a $200/month sub can allow.
Built by @regents_sh. Read more about the AI product lab here:
To join the success of autolaunch the ticker is already live on Base: $REGENT on Dexscreener ⟢