NEAR Intents Blur the Line Between What Humans and AI Can Accomplish

Whether it comes to solving advanced design problems almost as well as the experts, or taking on significant duties in the healthcare arena, AI agents have evolved very, very quickly.

Their rapid rise in use across industries, along with the verified accomplishments they’ve demonstrated, have shown that this could be the next big pivot for AI in general.

The Exponential Problem

After all, we are finding that large AI models can quickly be overwhelmed with too much data, too much processing, too much complexity. Taking on massive amounts of data and needing huge quantities of clean data just to train a model can be time-consuming, even with very expensive processors.

The problem is exponential: the more complex a model is, the resources needed to train it increase not linearly, but often exponentially.

One solution is to simply continue producing GPUs until we can brute-force the problem. But even if this is a good idea (it isn’t), there’s only so much additional performance we can extract from current-generation GPUs, even with novel approaches to AI algorithms.

Breaking Down Complexity

If we can’t brute force AI, then we can turn the problem on its side. This is what AI agents do, by breaking down a complex problem into a number of smaller, simpler problems. It doesn’t take much in the way of processing (or data) to solve a simpler problem. But solving a small problem on its own just gets us back to square one: limited intelligence.

The key to AI agents lies in our ability to make them significantly faster and cheaper than traditional AI models, while also enabling them to coordinate with other agents, humans, and an increasing number of online resources.

It’s that connective tissue we are so interested in when it comes to AI agents, because these simple problems can suddenly coordinate to solve increasingly complex challenges.

Growing Standards and Decentralization

There are growing standards when it comes to AI agents, giving them a more common structure and an ease with which they can interact with other agents, even those they’ve never encountered before, all without human help.

The final boost to the current growth of AI agent excitement is developing them to work in a decentralized ecosystem.

Blockchain: The Perfect Match

AI agents were not viable technology when blockchain was first developed, but you would be forgiven for thinking that the blockchain was designed specifically for AI agents.

It can operate without borders; it can transact in countless digital currencies, with wallets creating easy interfaces; it can interact with all types of counterparties (human or synthetic) without the need for trust, as smart contracts manage transactions with a decentralized and emotionless hand.

All of this has boosted the use and the potential for AI agents, but there is another element within Web3 that might be able to take AI agents even further, to the point that we may soon be interacting with agents for daily tasks, and not knowing, or caring, whether it is an agent or a human.

This element is the NEAR Protocol, which hosts key EVM compatibility with Aurora and its virtual chains, and NEAR Intents, a new paradigm for doing business. Let’s dive in to see just how much power these tools can give to AI agents, providing major benefits for the average user as a result.

NEAR Intents and AI Agents

First, a quick review of NEAR Intents. The protocol works across chains and creates a very broad marketplace for those who need something done. This “something” is specifically vague because there really aren’t limits to what could be proposed and delivered.

How It Works

The setup, at a high level, is straightforward. A user wants something done. This could be a transaction, a data fetch, an online transaction, and even more traditional services (the Docs mentions that pizza delivery isn’t out of the question).

This request is packaged up as an Intent, which describes what needs to be done in enough detail for potential takers to understand it, then estimate what cost/effort it would take them to fulfill the intent.

Once the intent has been completed and sent to the protocol, NEAR Intents presents it as a form of RFQ (Request for Quote) to the marketplace of potential providers. These providers could be humans, AI agents, or a combination of the two, depending on the Intent.

These providers, called Market Makers, will decide if they want to bid on the job, offer their price, and wait to see who wins the auction.

The user who created the Intent is presented with the offers, decides on a winner, and a contract is quickly agreed upon using fairly consistent terms to speed up the process.

The terms include what is expected, price, timing, and the process for conflict resolution. Once agreed to, the winning bid gets to work, completes the job, and gets paid.

A New Era of Commerce

This user-bid-market-quote-contract is certainly not new. This is essentially how humans have been conducting commerce from the very beginning. What is new is how it is done. Humans aren’t the only ones asking for work and providing quotes.

AI agents, who as a reminder are typically very good at solving narrowly defined tasks, are very well suited for a number of the Intents asked for by users.

Since AI agents can have considerable agency and intelligence, they can move between systems, make purchases, and conduct online work that can even extend to the physical world.

The pizza order example? A properly trained AI agent could do this, meaning that a deployed AI model could earn money by navigating a website or app, pay for a pizza on behalf of the requester, cause a real human to cook a pizza, and have another deliver it. All of this could be done autonomously with the architecture created by NEAR Intents and the rise of AI agents.

What Comes Next?

NEAR Protocol is certainly one of the key chains to watch. Through Aurora, the chain can launch virtual chains and interconnect across the ecosystem. Through NEAR Intents, an entire marketplace of buying and selling can happen, with limits we won’t even begin to approach for several years, and with amazing new use cases created by innovative users and providers.

The ecosystem is big enough to house the future factories for AI agents as well. Calyx, a NEAR ecosystem partner, has just launched funding for an advanced AI agent platform called Intellex, which focuses on improving AI agent interoperability far beyond what is currently possible.

We will see what the next few years bring, but it’s a smart bet that the news will be full of stories discussing how incredible AI agents have become.