1/6 @Mastercard just launched Agent Pay for Machines. New infrastructure built for AI agents to transact with each other, autonomously, at machine speed.
@Nevermined_ai is part of it 🧵
We still underestimate what AI agents could become in crypto.
A lot of people still think they're just smarter chatbots. They're not.
Chatbots are reactive. They answer questions.
AI agents are goal-driven.
Give them an objective, like researching a market, managing a portfolio, fixing code, or booking a service and they'll:
- plan the steps
- use the right tools
- evaluate the results
- adapt along the way
The difference is that crypto agents don't just analyse, they can act.
They can hold wallets, sign transactions, interact with DeFi protocols, monitor markets, post on social platforms, and pay for services, all with varying levels of autonomy.
They’re essentially large language models connected to wallets, blockchain infrastructure, APIs, and external tools.
Most crypto AI agents are built around a few core components:
• An LLM that handles reasoning and planning
• Tool integrations for web search, on-chain data, social media, APIs, and DeFi
• A non-custodial wallet to execute on-chain actions
• Memory that helps the agent stay consistent over time
In more advanced systems, multiple specialised agents work together: one planning, another researching, and another executing.
✦ Two ecosystems have emerged as early leaders:
- @virtuals_io Protocol on Base: focuses on creating tokenised AI agents that can own capital, perform jobs, and participate in agent-to-agent commerce.
- @elizaOS: has become one of the most widely used open-source frameworks for building AI agents that can operate across platforms like X, Discord, Telegram, and blockchain networks.
Another important trend is agentic payments.
AI agents can't easily open bank accounts or access traditional payment rails.
Stablecoins solve that problem by allowing agents to pay for compute, data, APIs, and even other agents' services through instant, programmable micropayments.
Projects & infrastructure working on this:
→ x402 (coinbase): Open standard for agentic/machine-to-machine payments
→ Google AP2: Agentic Payments Protocol
→ @moonpay agents: Non-custodial agent infrastructure
→ @crossmint: Agent wallets + virtual cards
→ @Nevermined_ai: Agentic payments + settlement platform
This is where crypto becomes a natural fit.
Blockchains give AI agents three things traditional systems struggle to provide:
- Native programmable money.
- Permissionless access to financial infrastructure.
- Verifiable identity and reputation through on-chain activity.
That's why many now see blockchain as the settlement layer for the emerging agent economy.
That said, we're still early.
Many so-called "AI agents" today are little more than chatbots with wallets attached. Security, reliability, and prompt injection remain real challenges.
The projects worth paying attention to are the ones generating actual usage, not just speculative token narratives.
Top projects on my watchlist 🔻
> $NEAR - @NEARProtocol: infra + settlement for the AI agent economy
> $SPECTRE - @Spectre__AI: AI market intel for crypto, stocks, on-chain
> $AEON - @AEON_Community: agent payments + real-world commerce
> $SERV - @openservai: reasoning + orchestration for prod agents
> $INTENT - @intentfitech: natural-language on-chain investing (swaps, yield, perps)
> $VEX - @ProjectVEXai: verifiable agent for on-chain capital + yield
> $FET - @Fetch_ai: decentralized network + marketplace for AI agents
> $TAO - @opentensor: incentivizing intelligency
✦ The bigger picture is becoming clearer:
As AI becomes increasingly autonomous, it will need a financial system that can operate without relying on traditional banks or constant human intervention.
Crypto provides that foundation.
The next generation of AI won't just answer questions. It will research, negotiate, transact, coordinate, and manage capital.
Crypto is giving those systems an economic layer to operate on.
Day 173 of 2026 building!
I built PocketAgent, a self-hosted payment delegation system for AI agents using the @Nevermined_ai's x402 Delegation Extension spec.
The idea: give your AI agent a "delegated card" with spending limits, auto top-up, and zero direct card exposure.
@Mastercard There are some things money couldn't buy.
Now your agents can. Metered and settled at machine speed, down to a fraction of a cent.
For the agent economy, there's Nevermined.
For everything else, there's Mastercard.
Proud to be a launch partner in Agent Pay for Machines.
Every year someone announces the agent economy, allegedly
Today Mastercard shipped actual settlement rails for it — cards, accounts, stablecoins — and @Nevermined_ai is building the layer that decides what agents buy and at what price.
Being right early is indistinguishable from being wrong, until a card network shows up
Backed @Nevermined_ai before agentic payments were consensus. Today the market is catching up.
Proud to see @dongossen and his team among the select partners building this next chapter of financial infrastructure alongside @Mastercard.
Early conviction. Exceptional founders. Category-defining outcomes.
6/6 Together, this is the foundation for an economy where agents are not just tools, they are economic participants.
We're early. But the infrastructure is being laid right now, and the companies building it today will define how the agent economy operates for the next decade.
Proud to be in the room where it's happening.
1/6 @Mastercard just launched Agent Pay for Machines. New infrastructure built for AI agents to transact with each other, autonomously, at machine speed.
@Nevermined_ai is part of it 🧵
5/6 @Mastercard brings the global network, the settlement rails, the compliance layer, and, perhaps most importantly, trust. @Nevermined_ai brings the commerce intelligence that sits on top.
Read full press release: https://t.co/va6QCQ3WdP