We are transitioning from a “play-to-earn” model to a more exciting era: games that are rich in genuine fun and infinitely scalable.
Author: Sid, IOSG Ventures
Original Title: IOSG Weekly Brief | The Integration of Gaming, AI Agents, and Cryptocurrency #260
Cover: Photo by Lorenzo Herrera on Unsplash
This article is for educational exchange only and does not constitute any investment advice. Please indicate the source when reprinting and contact the IOSG team for authorization and reprint instructions. All projects mentioned in the article are not recommendations or investment advice.
The Current State of Web3 Gaming
With the emergence of more innovative and attention-grabbing narratives, Web3 gaming as an industry has retreated in both primary and public market narratives. According to Delphi’s 2024 report on the gaming industry, the cumulative financing of Web3 gaming in the primary market is less than $1 billion. This is not necessarily a bad thing; it indicates that the bubble has burst, and current capital may be flowing towards higher quality games. The following figure is a clear indicator:
Throughout 2024, the number of users in gaming ecosystems like Ronin has surged significantly, and with the emergence of high-quality new games like Fableborn, it is almost comparable to the glory days of Axie in 2021.
The gaming ecosystems (L1s, L2s, RaaS) are increasingly resembling Web3’s Steam, controlling distribution within the ecosystem, which has become a motivation for game developers to create games within these ecosystems as it helps them acquire players. According to their previous reports, the user acquisition cost for Web3 games is about 70% higher than that for Web2 games.
Player Retention
Retaining players is as important as attracting them, if not more so. Although there is a lack of data on player retention rates in Web3 games, player retention is closely related to the concept of “Flow” (a term proposed by Hungarian psychologist Mihaly Csikszentmihalyi).
The “flow state” is a psychological concept in which players achieve a perfect balance between challenge and skill level. It’s like being “in the zone”—time seems to fly by, and you are completely immersed in the game.
Games that continuously create flow states tend to have higher retention rates for the following reasons:
Progressive Design
Early game: Simple challenges to build confidence
Mid-game: Gradually increasing difficulty
Late game: Complex challenges to master the game
As player skills improve, this meticulous adjustment of difficulty allows them to stay within their rhythm range.
Engagement Loops
Short-term: Immediate feedback (kills, points, rewards)
Mid-term: Level completion, daily tasks
Long-term: Character development, rankings
These nested loops can maintain player interest over different time frames.
Factors that disrupt the flow state include:
1. Improper difficulty/complexity settings: This may be due to poor game design, or even due to an insufficient number of players leading to match imbalance.
2. Unclear objectives: Game design factors.
3. Feedback delays: Game design and technical issues.
4. Intrusive monetization: Game design + product.
5. Technical issues/lag.
Symbiosis of Games and AI
AI agents can help players achieve this flow state. Before exploring how to achieve this, let’s first understand what types of agents are suitable for application in the gaming field:
The key to game AI is speed and scale. When using LLM-driven agents in games, each decision requires calling a massive language model. It’s like having an intermediary before every step you take. The intermediary is smart, but waiting for their response can make everything slow and painful. Now imagine doing this for hundreds of characters in a game; not only is it slow, but it’s also costly. This is the main reason we have not seen large-scale LLM agents in games yet. The largest experiment we currently see is the development of 1000 agents in Minecraft. If there are 100,000 concurrent agents on different maps, it would be extremely costly. Since adding each new agent leads to delays, players would also be affected by traffic interruptions. This disrupts the flow state.
Reinforcement Learning (RL) is a different approach. We think of it as training a dancer rather than giving step-by-step instructions through a headset. With reinforcement learning, you need to spend time teaching the AI how to “dance” and how to deal with different situations in the game. Once trained, the AI will make decisions naturally and smoothly in milliseconds without needing to request upwards. You can run hundreds of such trained agents in your game, each capable of making independent decisions based on their observations. They may not be as articulate or flexible as LLM agents, but they operate quickly and efficiently.
When you need these agents to work together, the real magic of RL shines. LLM agents require lengthy “conversations” to coordinate, while RL agents can form an implicit understanding during training—like a football team that has trained together for months. They learn to predict each other’s actions and coordinate naturally. While this is not perfect and they may occasionally make mistakes that LLMs wouldn’t, they can operate at a scale that LLMs cannot match. This trade-off always makes sense for game applications.
Agents as NPCs will address a core issue many games face today: player liquidity. P2E was the first experiment to use cryptoeconomics to solve player liquidity issues, and we all know how that turned out.
Pre-trained agents serve two purposes:
-
Fill the world in multiplayer games.
-
Maintain a level of difficulty for a group of players in the world, keeping them in a flow state.
While this seems very obvious, it is challenging to build. Indie games and early Web3 games often lack the financial resources to hire AI teams, which provides opportunities for any service provider with an RL-based agent framework.
Games can collaborate with these service providers during trial and testing phases to lay the foundation for player liquidity at launch.
This way, game developers can focus their main efforts on game mechanics, making their games more enjoyable. While we love to integrate tokens into games, games are ultimately games, and they should be fun.
Agent Players
The Return of the Metaverse?
One of the most played games in the world, League of Legends, has a black market where players train their characters with the best attributes, which the game prohibits.
This helps form the basis for game characters and attributes as NFTs, thus creating a market for this.
What if a new subset of “players” emerges as coaches for these AI agents? Players could guide these AI agents and monetize them in different forms, such as winning competitions, and could also serve as “training partners” for esports players or passionate gamers.
The early version of the metaverse may have just created another reality instead of the ideal one, thus failing to meet its goals. AI agents help metaverse residents create an ideal world—an escape.
In my view, this is where LLM-based agents can shine. Perhaps someone can join pre-trained agents in their world, all of which are domain experts capable of conversing about their favorite things. If I create an agent trained on 1000 hours of interviews with Elon Musk, and a user wants to use an instance of this agent in their world, I could be rewarded for that. This could create a new economy.
With metaverse games like Nifty Island, this could become a reality.
In Today: The Game, the team has created an LLM-based AI agent named “Limbo,” with the vision of multiple agents interacting autonomously in this world while we can watch a 24×7 live stream.
How Does Crypto Integrate?
Crypto can help solve these problems in various ways:
-
Players contribute their game data to improve models, gain better experiences, and thus receive rewards.
-
Coordinate multiple stakeholders like character designers, trainers, etc., to create the best in-game agents.
-
Create a market for ownership of in-game agents and monetize them.
There is a team doing all these things and more: ARC Agents. They are addressing all the issues mentioned above.
They have the ARC SDK, which allows game developers to create human-like AI agents based on game parameters. With very simple integration, it resolves player liquidity issues, cleans up game data and turns it into insights, and helps players maintain flow states in games by adjusting difficulty levels. For this, they use Reinforcement Learning (RL) technology.
They initially developed a game called “AI Arena,” where you basically train your AI characters to fight. This helped them form a benchmark learning model that constitutes the foundation of the ARC SDK. This creates a sort of DePIN-like flywheel:
The Chain of Thought team has explained this well in their article about ARC agents:
Games like Bounty are taking an agent-first approach, building agents from scratch in a wild west world.
Conclusion
The integration of AI agents, game design, and Crypto is not just another technological trend; it has the potential to solve various issues that plague indie games. The brilliance of AI agents in gaming lies in their enhancement of what makes games enjoyable—healthy competition, rich interaction, and captivating challenges. As frameworks like ARC agents mature and more games integrate AI agents, we are likely to see entirely new gaming experiences emerge. Imagine a world that is vibrant not because of the presence of other players, but because the agents can learn and evolve with the community.
We are transitioning from a “play-to-earn” model to a more exciting era: games that are rich in genuine fun and infinitely scalable. For developers, players, and investors focused on this field, the coming years will be thrilling. Games in 2025 and beyond will not only be technologically advanced but fundamentally more engaging, accessible, and vibrant than any games we have seen before.
Disclaimer: As a blockchain information platform, the articles published here only represent the personal views of the authors and guests and do not reflect the stance of Web3Caff. The information in the articles is for reference only and does not constitute any investment advice or offer, and please comply with the relevant laws and regulations of your country or region.
☞ Explore Web3 industry innovations from a deep perspective by subscribing to the official WeChat account of Web3Caff ↓