5 Amazing Use Cases for Based Agents and Near AI Assistant

Imagine building an AI agent on Base that can track all your cryptocurrency transactions and automatically earn you enough to fully manage your USD tax liabilities.

Or an agent that will automatically rebalance your cryptocurrency portfolio so that your profits automatically go into quality coins like Bitcoin, Ethereum, Solana and you don’t end up with 90% shitcoins that go back to 0 at the end of the transaction. to give.

Magazine imagined these things and pitched them to Nemil Dalal, Senior Director of the Coinbase Developer Platform, in hopes that someone could build them. Base is creating the necessary AI tools so that anyone with a brilliant idea can create one, regardless of technical ability.

The platform is targeting one million developers built on Base, and this will only be realized with the help of AI coding.

“We’re only scratching the surface of how much productivity we can gain, but within six to 12 months, we can see really fantastic apps being built with AI with just a few types of prompts,” he says. .

“I think we live in a magical age of architecture. “I think software engineering will change more in the next year than it has in the last 30 years.”

As it happens, Dalal really likes the idea of ​​automatic portfolio rebalancing.

“I think this is another really good example,” Dalal said, sitting across the table from a McDonalds in Bangkok. There we had no choice but to retreat from Base’s noisy “Every AI Agent Needs a Wallet” nearby event.

“I think it’s a really good idea to be able to put it out there on paper as an example of what people can actually create,” he says.

AgentKit on the Coinbase Developer Platform

The Coinbase developer platform isn’t yet available for retail, but the AgentKit infrastructure seems easy enough for the average person to use. Going back to the event, there is a demo where a user logs in and uses a cryptocurrency wallet to spin up the underlying agent and then perform a transaction. The whole thing takes about 1 minute.

It currently uses OpenAI’s API keys, but is model agnostic and will work with Claude or open source models as well.

One of the projects built on Base is Brian Knows AI (no relation to Brian Armstrong). By creating a natural language voice interface, you can ask AI to perform complex cryptographic tasks and the AI ​​will perform them.

“You can basically talk to your wallet in natural language,” explains Dalal. “For example, people may not want to use MetaMask. They may want to be able to easily send money using natural language.”

It’s easy to see the potential for that technology to completely transform the UX for cryptocurrencies, allowing users to enter the space without knowing anything about their private keys and unreadable wallet addresses.

Base’s motto is to build 1 billion users and 1 million developers. Senior contributor Jesse Pollak explains to the Magazine that keeping buildings simple is key to L2’s growth strategy.

“From a technical standpoint, we’ve helped create a set of tools that make it very easy for people to build, with on-chain kits, smart wallets, all the new AI tools that allow you to build agents, etc. Easy, easy, easy to build and create in Base.”

Warren Buffett Agent Building

Another cool idea discussed by the Base team is training your agents on the strategies of real investment professionals and having them invest on your behalf.

“We joke about it, but one of the things we’re really excited about is being able to basically educate someone on all the writing, like Warren Buffett, and then interact with that personality.”

Given that Buffett famously labeled Bitcoin “rat poison squared,” he may not be the best choice for cryptocurrency investors.

“In Warren Buffett’s case, he probably won’t be sending crypto or NFTs anytime soon. But this is an example of what we can do right now,” Dalal laughs.

The user-built ecosystem is Base’s growth strategy.

Both Pollak and Dalal are very passionate and excited about the future of cryptocurrency. Considering the financial nihilism infecting the universe and the apocalyptic fears of artificial superintelligence, it’s truly surprising to meet an idealist who can only see the positives at the intersection of AI and cryptocurrency.

Also read: A bizarre cult is growing around the AI-created memecoin ‘religion’.

Dalal explains that underlying agents began in earnest about five months ago, when Coinbase began building the technology needed to give agents’ cryptocurrency wallets the ability to trade autonomously.

“We saw what was happening with OpenAI, but none of them were on-chain. “The agents were all independent of the blockchain,” he said. Since agents cannot obtain bank accounts, the mission has become to overcome technical obstacles that prevent agents from using cryptocurrency wallets, he added.

“So we started building and basically had a slogan that said every AI agent deserves a wallet.”

AI agents and the hallucination problem

Agents might search the web, post to What if we harbor the illusion that Tron is a good project?

Dalal says this is why Base initially focused on fun, low-value use cases like flipping memecoins or issuing NFTs.

“These are just for fun. So you don’t need a lot of money for AI to be able to move on its own,” he says.

“The second is that, at least right now, AI agents have to have humans in the loop.” He explains that over time, more granular authorization will be enabled, so users can specify that agents can spend up to $10 per 100 times, but anything beyond that will require human approval.

Virtual Protocol and LUNA

One of Base’s most famous AI agents is Luna, a virtual social media influencer and singer. Luna started automatically tipping users to engage with her content. Retail users can spin up agents or monetize them through the virtual protocol, like Pump.fun launching memecoin, and agents will be released from X once the market capitalization of the associated token reaches $213,000. Token holders participate in decisions about the agent’s actions. Develop and share generated profits.

It’s still early days, and Luna has a market capitalization of about $61.2 million. One of the most popular agents on the platform is AIXBT. AIXBT analyzes cryptocurrency X sentiment to identify trends. “There was a clear path to 100 million mcap based on current memecoin dynamics,” he said on November 3, calling memecoin PNUT. “Enter now before the norm is established.” This was good advice, considering PNUT’s current market capitalization is $1.8 billion.

The future of AI is near

Base is not the only blockchain that uses AI agents and cryptocurrency wallets. Near Protocol has released an AI agent in alpha called AI Assistant that leverages chain abstraction and intent to enable cross-chain crypto exchanges and payments. In addition to making the cryptocurrency UX more seamless, cross-currency swaps are also a necessary step to enable AI agents to hire and pay agents from other networks or Web2 to solve specific problems.

Co-founder Illia Pollushkin said the company will soon add the ability for assistants to accept fiat payments on Web2 services, such as booking airline tickets or ordering takeout. On the backend, intents work by connecting a user with another party that executes the action the user wants to perform.

“You say, ‘I want to exchange this token for this token,’ right? On the other hand, someone says, ‘Okay, let me guess this, I’m actually going to be the other person.’

Enabling agents to accept payments in fiat currency works in a similar way, but requires a few additional steps.

“Someone can say, ‘Okay, I’m going to pay fiat into this account.’ Now fiat currency is encrypted and not secure. So we need some form of dispute resolution,” he explains.

There are also compliance issues around KYC or terms of service consent that need to be communicated back to the user.

“This is the framework we are introducing to connect AI to blockchain and the real world. Any service, any market maker, any AI agent can log into the other side and say, ‘Hello, I’ll do this.’ Yes, you have the correct proof. I agree with the dispute method.’ So we effectively created this marketplace between user requests and responses from various participants.”

AI turns everyone into a builder.

Both Near and Base are excited about the possibility that anyone with a good idea can build a DApp using AI agents, even if they can’t code. Currently, there are only about 30,000 to 40,000 developers building on-chain, and while Near has set a target of 10,000 user-built DApps, Base wants to have 1 million developers building on-chain.

The audience for Every AI Agent Needs a Wallet was noticeably diverse, and Pollak had just returned from a world tour, visiting several countries to learn about local needs and wants for Base. Dalal said AI tools will make the platform much more accessible to people around the world who previously had no training or resources to build apps.

“So anyone who has access to this model can build it. This model is relatively affordable to any builder around the world today. “This is the really powerful part,” he says.

subscribe

The most interesting read on blockchain. Delivered once a week.

Andrew Fenton

Andrew Fenton, based in Melbourne, is a journalist and editor covering cryptocurrency and blockchain. He has worked as a national entertainment writer for News Corp Australia, a film journalist for SA Weekend and The Melbourne Weekly.

Follow the author @andrewfenton