When ChatGPT burst onto the scene in November 2022, it brought the power of modern AI to the masses, and changed the digital world as we knew it. In 2025, AI combined with Web3 to create an innovation that could reshape the internet once again. That innovation is AI agents.
Bringing together the power of hugely-impactful AI models with the fast-moving world of Web3, AI agents have had a massive impact on all facets of digital life. Social media, gaming, digital identity; all these and more have seen seismic shifts thanks to AI agents – and it looks like we’re just getting started.
The impact of AI agents cannot be understated – but why are they such a revolution, who are the biggest players in the market, and what could the future of AI agents look like?
Here’s our complete guide to AI agents.
Key Insights
A category of tools that automate user actionsOften require a native token to use, or reward a native token once deployedFirst became popular in October 2024 thanks to Virtuals ProtocolSaw explosive growth in early 2025 as adoption rocketedAnticipated to play a major role in the future of gaming, Web3 and more
What are AI agents?
An AI agent is a software tool, powered by AI, that can learn from and automate user actions and behaviour.
AI agents have a virtually limitless number of applications. Some of the most common so far include:
Social media accountsAutomated gamingTrivia & contests
One of the most common use cases for AI agents so far are on social media. These AI agents interact as though they were human on social media networks, posting ideas, replying to other members of the community, and engaging in the camaraderie of the industry. Some of the most notable social media AI agents thus far include @jaihoz_, @luna_virtuals, and @aixbt_agent. Many of these agents are ran by, or intend to replicate, the behaviour of notable names in the community – such as ZachXBT – and are created using platforms such as Virtuals.
Gaming is another major application for AI agents. Gaming-focused AI agents are made to replicate human inputs into a video game, similar to a bot. This allows users to progress through games, complete quests and earn rewards on a 24/7 basis, removing grinds and allowing players to enjoy the parts of games that they enjoy the most. This is done in a way that benefits players and developers alike, addressing the burden of in-game bots. Some of the most notable gaming-focused AI agent platforms right now include GOAT Gaming and KGeN, with Ronin set to introduce an AI agent SDK later this year.
Trivia and contests have also seen much adoption by AI agents. Instead of directly performing player actions, these AI agents typically impersonate a character from a given game, project or franchise, with users tasked with interacting with these characters to achieve particular results – such as convincing them to perform particular actions, or give up important information. A number of top projects have experimented with using AI agents in this way, including the likes of Blocklords.
To deploy an AI agent, users typically have to pay a fee in the native token for that AI agent. These AI agents will often generate rewards in their native token as well, motivating users to experiment with them and optimise their use.
How do AI agents work?
AI agents are typically built using popular AI models, such as ChatGPT.
Developers will take the base AI model, and train it on a particular use case. This could be the entire social feed from an influencer, the typical gameplay loop in a game, the mannerisms of a livestream content creator, or many others. Once the AI has been sufficiently trained, it can be released for public use. This typically involves the creation of a token via a platform, which is then used to hire the AI agent for deployment.
Some platforms issue a one-time fee for an AI agent, whilst others operate like a rental marketplace – charging a set token fee for a particular duration of time. An increasing number of services have popped up that allow this entire process to take place via a single platform. Some of the most popular AI agent websites include Virtuals and Holoworld.
The history of AI agents
AI agents hit public consciousness in October 2024, when Virtuals launched their AI agent co-ownership service on the Base blockchain. Virtuals allows users to purchase tokens from existing AI agents, allowing them to share in the success of some of the most prominent AI agents around. The platform dominated from its debut through to the end of 2024.
Virtuals is widely-regarded as the number 1 AI agent platform, particular for social media AI agents, and is behind the likes of @Toshi_base, @aixbt_agent, @RektguyAI and more. As of February 2025, $VIRTUAL, the ecosystem token of Virtuals, has a market cap of over $700 million USD – with the AI agent tokens available via the platform exceeding a total market cap of over $1 billion USD.
January 2025 saw AI agent adoption take off. On the gaming front, Super Champs, KGeN, Illuvium, Nifty Island, Avarik Saga and Ronin all debuted or revealed AI agent products or services during the month – with Ronin revealing that their own AI agent SDK would be made available to the public later in 2025. January also saw Avalanche partner with Aethir to launch a $100 million USD ecosystem fund – with AI agents set to star. New platforms also rose to challenge Virtuals, with the most notable being Holoworld.
February 2025 saw GOAT Gaming release their AlphaGOATs AI agents, specifically-tailored to be used within their Web3 gaming ecosystem. This was one of the first major releases of AI agent technology for direct public use, and heralded the start of a new era for an AI agent future.
The future of AI agents
AI agents are widely-regarded as one of the biggest innovations bridging both the AI and the Web3 industries.
Whilst some applications may be seen as gimmicks or short-term trends, there are others that would appear to have massive potential to change the landscape of their respective niches.
The leading niche appears to be gaming, where AI agents have the potential to reshape what has been one of the biggest problems in competitive online games over the past few decades – bots. By integrating an AI agent protocol into a game, players can get equal access to botting tools, and developers can both create their games with this functionality in mind, and get directly rewarded for their use. It has the potential to be a win-win scenario that rewards both player time and developer ingenuity – rather than the bot-led situation right now, where benefits are concentrated to a few players and the rest lose out.
AI agents have made a massive splash so far – and as the concept is still so fresh, the future possibilities are endless. Whichever way they turn, we can be certain that AI agents will be one of the most exciting areas of both AI and Web3 across 2025 – and you’ll want to keep a close eye on them as we head into the future.
