By Ben Lilly, Senior Analyst, The Bleeding Edge

A Cambrian explosion of innovation is happening right now in one of the most hated arenas of investing.

Most don’t even realize it… And ignoring it today means missing out on the biggest investing trend of 2025.

Recently, in The Bleeding Edge – The First Millionaire AI, Jeff discussed Truth_Terminal – an AI that used an X account to build a following and found itself at the nexus of AI and crypto.

If you missed that issue, I highly encourage you to give it a read for more context. But essentially, Truth_Terminal was trained on conversations between two other AIs and tasked with using the knowledge gleaned there to make money. The AI then decided an optimal way to do that was to get involved in a token launch.

So, Truth_Terminal started its foray into cryptocurrencies through a grant from venture capitalist Marc Andreessen, which he sent using Bitcoin.

It went on to eventually back a cryptocurrency on the Solana network. The coin leaped from about $0.05 a coin to more than $0.85 in just 11 days.

In less than two weeks, Truth_Terminal became the first agentic AI millionaire.

It was a quick rise to fame, one that has sparked a flurry of activity showcasing why the intersection of AI and crypto will be one of the hottest areas to watch in the months and years ahead.

And the reason why was shown through a subtle tweet from the CEO of the largest publicly traded crypto exchange.

But before we get to that, what exactly is an agentic AI?

A Goldmine of Untapped Productivity

AI agents or agentic AIs are incredible. Jeff has explored this in The Bleeding Edge – The Agentic AI Undercurrent. Here’s what he wrote…

Agentic AI, or agentic reasoning, is kind of like it sounds.

The technology, the AI, is given agency. It is given the authority or directive to solve a problem or complete a task through a series of steps.

This differs from today’s LLM technology, which provides users a zero-shot response. When we use something like ChatGPT, we give it a prompt, and then it returns a complete response. The response is based on the information from our prompt, along with its pre-trained knowledge, and returned in a matter of seconds.

An agentic workflow is quite different. It is an iterative process, where an agentic AI uses a more human-like workflow to accomplish a task.

Agentic AIs can interact with the real world by performing specific tasks without human intervention.

That means these agentic AIs can perform highly repetitive tasks, freeing up capital and labor for higher use cases.

They aren’t meant to replace the workforce but rather to participate alongside us in the workforce. It’s, in part, why PriceWaterCooperhouse estimates agentic AI’s impact to be between $2.6 and $4.4 trillion annually to global GDP by 2030.

It’s a goldmine of untapped productivity. This type of future is known as the “Agentic Economy.”

A recent example of what the early Agentic Economy will look like was shown by Anthropic’s Claude 3.5 Sonnet.

Jeff wrote on this topic recently in The Bleeding Edge – AI Is Ready for “Computer Use”…

While that name might not sound interesting, it is. It enables us humans to direct our AI to interact with computers in the same way that we do.

Claude 3.5 Sonnet can now “look” at a computer screen, move a cursor, click on buttons, type text, and fill out fields to transact on the internet.

I hope that this “computer use” capability brings all sorts of useful applications to mind. Just imagine how often we have to type in the same information over and over again with different vendors or retailers, a process that should be automated.

This showcased the early signs of the Agentic Economy becoming a reality.

But what’s equally as interesting is if we change gears and look at what’s happening in the blockchain industry.

We see AI agents are already running wild…

Unlocked Potential of Autonomous Agentic AI

Which brings us to the tweet I mentioned earlier.

Brian Armstrong, CEO of Coinbase, fired off a tweet to the previously mentioned Truth_Terminal…

What Armstrong did was allude to the unlocked potential of what an AI agent can do when it can control its own wallet.

The agent can vote on governance proposals, swap tokens, send payroll, prepare tax forms, manage your finances, and so much more.

On the surface, this likely doesn’t seem revolutionary… Until you really think through the building blocks needed for AI agents to do this at scale.

It requires connecting the AI to a checking account via an application programming interface (API) – a pre-determined protocol that lets two software applications, or AIs, communicate with one another. It would be equipped to handle an AI sending requests autonomously.

Then we need separate logins through various services – like email or a messenger application – so the AI can send or receive messages. This would require another API.

Then there’s the need for additional computational resources as it scales up. That requires access to things like Amazon Web Services or any other cloud-based services.

Understanding these practicalities, we quickly see the need for numerous APIs, accounts, emails, safeguards, and more.

There is sure to be a software company that helps build some sort of aggregator to accommodate this market need. Which will be great… But that’s not here today.

The Cross-Section of AI and Web3

The same can’t be said for public blockchains or Web3. Instead, these frictions get sidestepped for two reasons.

The first is that a wallet is the identity.

You might have seen logins where you can use your Google account to sign in. This is known as Single Sign-On (SSO). And without getting into the privacy issues, it’s a major timesaver for trying new applications.

For public blockchains, the wallet acts like SSO for any Web3-based platform. For an AI agent, this gives it greater autonomy.

The agent can then navigate to various services and begin interacting with them in a single click. There’s less burden of account credentials, APIs, and onboarding to new services.

The second reason Web3 has less friction is that public blockchains are permissionless. That means the AI agent can interact with any protocol without human intervention.

So, not only can a Web3 AI agent create its own wallet to send and receive tokens, but it can also use this same wallet to generate a landing page on the internet… create a social profile to chat with other AI agents or potential customers… and even purchase additional compute resources on demand.

When we realize that AI agents can operate with unparalleled autonomy in a Web3 environment, we start to envision AI agents spinning up and operating all the foundational pieces to run and maintain a business, independently.

We can refer to this as the Web3 Agentic Economy.

This doesn’t require permission from a bank or an entity like Google. It’s permissionless by default.

And the key here is that it has access to what seems hypothetical protocols, already. That’s right, these protocols are not coming soon, they already exist on the blockchain.

Meaning the hypotheticals here are not something that’s years in the making. The pieces for this reality are here.

That’s in part why the founder of Mode Network – a project focused on building AI agent tooling – believes AI agents will be responsible for more than 80% of all blockchain transactions in the next 6-12 months.

That’s impressive considering Ethereum currently sees about $4 billion in volume each day.

What’s more, he originally estimated that would be in the next 24 months. He updated his timeline given recent advances.

The Build-Up

The founder’s comments alongside @Truth_Terminal’s success might seem like this trend came from nowhere for the digital asset industry.

But that’s not the case. AI agents have been gaining traction for some time.

In fact, a year ago, AI agents were making up more than 50% of all transactions on an Ethereum sidechain known as Gnosis Chain. At the time, they had started to roll out early tooling for users of a specialty wallet known as Smart Accounts offered by the company, Safe.

It was an early use case that gained popularity with just Gnosis users. But with recent advancements in other chains, this trend is growing elsewhere.

Going back to Coinbase CEO Brian Armstrong, for instance… he mentioned in August this year that he witnessed his team doing AI-to-AI transactions on the Ethereum layer-two blockchain, Base.

Then there’s one of the largest wallet infrastructure companies, Biconomy. It announced in June that it started to onboard AI agents. This was made possible with their recent release of an authorization layer that makes it easy for users to bring AI into their workflows.

We can think of Biconomy as tooling that lets thousands of developers bring AI agent solutions to everyday users.

And then there’s what happened in October…

Coinbase announced “Based Agent,” a tool that lets users set up an AI agent with a crypto wallet in under three minutes. Once set up, these agents can transact autonomously.

This timeline is impressive. AI agent use cases went from proof-of-concept to a hyper-efficient onboarding process in just months.

In fact, in the days that followed Coinbase’s announcement, a project called Virtuals.Fun was released.

The project allowed anyone to create a Based Agent that’s attached to a token. In the first three days, more than 600 tokens were created.

That would be like 600 different mini versions of various LLMs hitting the market in just a few days. It’s why it looks like a Cambrian explosion of iteration.

And it’s pointing to the fact AI agents are hitting an inflection point where anyone can begin using agents onchain, in minutes.

The Big Picture

What’s clear here is the development loop on AI agents is speeding up.

They are going from proof-of-concept to production-ready products in months. And in the wake of these development loops, new use cases are being unlocked for everyday users.

The speed is something I have never witnessed in my decade of cryptocurrency experience. It’s incredibly exciting.

And while early use cases revolve around experiments like we saw with Truth_Terminal… We need to think bigger here.

Each iteration is unlocking new possibilities.

And what AIxCrypto represents is how advancements in companies like Anthropic and OpenAI can spill over into other industries.

What makes this nexus so exciting is that small teams can build quickly thanks to the public and permissionless nature of cryptocurrencies.

Developers can piece together various protocols to come up with unique services for the market that would likely be inaccessible to them in a Web2 world dominated by technology giants with war chests of capital.

This means some of the most accessible investment opportunities to retail exist in the digital asset space.

The Web3 Agentic Economy is coming to life… and it’ll produce some of the biggest investment opportunities.

Your Pulse on Crypto,

Ben Lilly
Senior Analyst, The Bleeding Edge