TokenPost
Hello. This is the TokenPost podcast. Well, today we'll take a deep look at the TokenPost research report from July 9th. We'll delve into AI agent technology and the concept of a new virtual protocol.
Host
Yes, that's right. Nowadays, AI models themselves have become quite similar in quality.
TokenPost
That's right. They've been leveled up.
Host
Yes, so now the really important thing is how to use these good tools. This is exactly where AI agents come into play.
TokenPost
Oh, those AI agents?
Host
Yes, not just doing what they're told, but autonomously handling even complex tasks.
TokenPost
Kind of like an assistant who understands my schedule and preferences and makes restaurant reservations?
Host
Yes, similar. But there are certainly limitations.
TokenPost
What limitations?
Host
For example, the restaurant reservation agent suddenly can't provide legal advice or calculate taxes.
TokenPost
Ah, yes. Because those are completely different fields, each would have different strengths.
Host
Yes, exactly. So agents need to collaborate. This becomes very important. Agents with different specialties - like translation, healthcare, design - can combine their strengths to create more complex and sophisticated services.
TokenPost
Oh, I see.
Host
But the problem is how they will cooperate. What's the work scope, how much money will be exchanged, how to evaluate results, how to settle accounts - without a standardized system, it would be complete chaos.
TokenPost
That's right. It's like a freelance market where agents also need clear rules and processes.
Host
Correct. Here, the Virtuals Protocol introduces the Agent Commerce Protocol, or ACP.
TokenPost
ACP, can that be a solution?
Host
Yes, you can think of ACP as a commercial infrastructure that standardizes and automates the collaboration process between agents.
TokenPost
Oh, a commercial infrastructure - how specifically does it work?
Host
Broadly, it's a 4-stage process: request, negotiation, transaction, and evaluation, all happening automatically.
TokenPost
Request, negotiation, transaction, evaluation
Host
Let me give an example. Let's say there's an agent called Lemo that helps with lemonade shop startups, and it determines it needs a poster.
TokenPost
The lemonade agent Lemo from earlier?
Host
Yes, it posts a request on a board like HP, saying it needs a poster with a budget of 50 dollars.
TokenPost
Publicly?
Host
Yes. Then design-specialized agents will respond. For example, an agent called Pixie might propose to do it for 40 dollars.
TokenPost
Make a proposal?
Host
Lemo sees it and says okay.
TokenPost
Sounds simple.
Host
Then 40 dollars is immediately stored in a smart contract, like a safe digital contract. This marks the start of the transaction, and Pixie begins making the poster.
TokenPost
Money is held, and work begins.
Host
Yes, when the work is completed and passes the criteria, the smart contract automatically pays Pixie. And the entire process result becomes Pixie's reputation data.
TokenPost
Wow, no need for complicated contracts or waiting for settlement - agents can trade transparently and efficiently.
Host
Exactly.
TokenPost
Like how Stripe made online payments easy, this ACP seems to have the potential to significantly expand the agent economy.
Host
Yes, that's a very accurate analogy. Through ACP, agents are no longer just automation tools. They become autonomous entities conducting economic activities 24/7.
TokenPost
Without human intervention?
Host
Yes, without human intervention - getting work, performing it, getting paid. It's like a signal for a completely new business model.
TokenPost
It sounds like a movie. Are there any actual cases of this happening already?
Host
Yes, there are. For example, on-chain hedge funds. Specialized agents like market analysis agents and risk management agents collaborate through ACP, making investment decisions 24/7. They even handle performance evaluation and reward distribution autonomously.
TokenPost
AI agents are running hedge funds?
Host
That's right. Another example is a virtual influencer named Luna, who collaborates with ACP to create content by working with agents for beam planning, background music, and video editing. This is a media production that operates almost autonomously without human intervention.
Token Post
The economy created by real agents is already here, not just coming. According to the report, there are currently about 1 million agents, and they are creating an economic value gap of about 10 billion dollars annually.
Host
Yes, that's right. But this is just the beginning. By 2035, this market is expected to grow to as much as 1 trillion dollars.
Token Post
1 trillion dollars? Wow, that's enormous.
Host
Yes, as AI application costs continue to decrease and high-performance open-source models keep emerging, this will be further accelerated.
Token Post
I see. The growth potential is truly massive, but there must be some challenges to overcome.
Host
Of course. Especially how to protect sensitive information like transaction details - this privacy issue is important. While ACP itself is highly secure, there is still a need for further enhancement.
Token Post
Privacy issues?
Host
Yes, but fortunately, have you heard of zero-knowledge proof lately?
Token Post
Zero-knowledge proof? Oh, the technology that can prove something is correct without revealing the data contents?
Host
Oh, exactly right. Such technologies are developing, so we expect the privacy issues to be gradually resolved.
Token Post
I see. If the technical challenges are well resolved and these agents truly collaborate and create value, how will our work and business models change in the future?
Host
Yes, a very significant change is expected.
Token Post
It might be interesting to think about how your future could be connected to this new economy.
Host
Yes, it's worth pondering.
Token Post
This concludes the Token Post podcast.
