Gaming x AI agents: an emerging trend for 2025?

Gaming: A Colossal Industry

Far from being a niche market, gaming is now a colossal sector and one of the most dynamic cultural industries worldwide, with no signs of slowing down. The total market capitalization of the video game industry is estimated to exceed $1 trillion by 2025.

In 2024, the global video game market generated an estimated $187.7 billion in revenue, marking a 2.1% increase from the previous year. Some analysts project a threefold increase by 2035. To put this into perspective, the gaming industry already generates twice the revenue of the film and music industries combined.

This growth is supported by a continuously expanding player base. In 2024, there were over 3 billion gamers worldwide across all platforms, from consoles to mobile games. In France, video gaming is a common activity for 70% of the population, with a nearly equal distribution between men (51%) and women (49%). Casual gamers have an average age of 39 and spend approximately six hours per week playing.

---

AI in Gaming

Another industry that has been booming in recent years is Artificial Intelligence (AI). The AI market accounted for $184 billion in revenue in 2024, with projections exceeding $500 billion by 2028. What would happen if these two industries merged?

The gaming industry did not wait for the emergence of ChatGPT to integrate AI—on the contrary, AI is already ubiquitous in video games. For decades, AI has been used to animate NPCs (Non-Player Characters), create dynamic game environments, and adjust difficulty levels based on player performance.

• Autonomous and realistic NPCs: AI enables the creation of characters that interact with players more credibly. In Red Dead Redemption 2, NPCs have routines and memory, influencing their behavior and interactions with players. In the future, integrating advanced AI agents could allow NPCs to have complete autonomy, capable of exchanging items, making decisions, and evolving in a persistent world.
• Procedural content generation: AI is already used to create maps, quests, and dynamic environments. In Minecraft, each new game generates a unique world, while No Man's Sky relies on AI to generate infinite new planets.
• Difficulty adaptation: Some games adjust difficulty levels based on player performance. Resident Evil 4 modulates enemy aggression in real-time, ensuring a smooth and engaging experience.
• Graphic and performance optimization: Solutions like NVIDIA DLSS leverage AI to enhance image quality in real-time while reducing hardware resource consumption.
• Matchmaking and cheat detection: In competitive games like League of Legends, AI is used to balance teams and detect suspicious behavior to combat cheating.

The examples above are just a few illustrations of AI's various advancements in gaming over the years. However, the real revolution is emerging with the arrival of autonomous AI agents in video games. These agents could actively participate in game economies and interact directly with the blockchain, drastically transforming the user experience.

→ For more on the subject, find the full video on Victor's YouTube channel

---

Synergies Between AI, Gaming, and Web3: AI Agents

The integration of Web3 into gaming has often been associated with NFTs and cryptocurrencies, with mixed results. However, a new emerging vertical is the introduction of AI agents on the blockchain, capable of owning and exchanging digital assets without human intervention.

An AI agent is a program capable of executing actions autonomously to achieve a specific goal. In the blockchain context, these agents can hold their own wallets, interact with smart contracts, and perform actions without human supervision to reach a given objective.

→ Read our full presentation on AI agents on the blockchain:

In gaming, these AI agents could take the form of evolving NPCs in various formats: an enemy capable of making more advanced decisions based on the player's strategy in a strategy game or a merchant with whom players can negotiate beyond predefined scripts.

Web3 integration brings three major improvements to AI agent development in gaming:

• Autonomous wallets for AI agents: Thanks to smart contracts, an AI agent can hold, buy, and sell digital assets associated with the game's economy (tokens, NFTs).
• Interaction without intermediaries: Unlike Web2 games, where everything is controlled by the publisher, a Web3 game can allow AI agents to function entirely independently.
• An interoperable exchange ecosystem: An AI agent trained in one game could be used in another while retaining its history and abilities.

Here are some concrete use cases:

• In a Web3 MMORPG, AI agents could be autonomous merchants, negotiating with players and adjusting their prices based on supply and demand.
• Players could train and customize their own AI agents, improving their skills to resell them on a secondary market as NFTs.
• Quests and events could be generated in real-time, with AI agents reacting to player choices and creating unique scenarios.

In practice, this would enable the development of a new form of economy within a video game, where NPCs (AI agents) play an integral role. This could create new game mechanics and new forms of interactions and cooperation between players and NPCs. Overall, AI agents have the potential to drastically enhance NPC performance in video games and create entirely personalized gaming experiences.

---

Mapping the Gaming x AI Agents Sector

The rise of AI agents in the Web3 ecosystem has given birth to numerous projects exploring this technological convergence. These projects are structured around several categories: frameworks and launchpads for AI agents, specialized blockchains, infrastructure and SDKs, AI-driven metaverses, as well as Web3 games integrating AI agents. This mapping provides an overview of the key players shaping this emerging trend.

Frameworks and Launchpads for AI Agents

The emergence of AI agents in Web3 gaming is accompanied by the development of dedicated frameworks and launchpads. These tools enable developers to design, train, and deploy AI capable of interacting with game environments. Some offer low-code or no-code solutions, facilitating the integration of autonomous agents without requiring advanced AI expertise.

• Virtuals

Virtuals has established itself as the leading launchpad for AI agents. Its GAME framework allows defining an AI agent’s personality, optimizing its decision-making process, and enabling interactions in various environments, including virtual worlds. An experiment was already conducted on Roblox, where an AI agent created with GAME was left to operate autonomously to observe its behavior in a simulated environment.

• Daydreams

Developed by a key player in the Starknet ecosystem, Daydreams is a generalist framework allowing the creation of advanced AI agents. Unlike solutions focused solely on gaming, Daydreams also offers functionalities for autonomous on-chain trading, on-chain data analysis, and AI agents capable of reasoning, planning, and executing complex actions. The project's ambition is to provide AI capable of contextualizing their environment and making smarter decisions.

• ai16z and the ElizaOS Framework

Originally designed to create influencer AI agents interacting on various platforms, the ElizaOS framework developed by ai16z could evolve with its upcoming V2, expected in the next few months. A synergy with Daydreams could allow Eliza to be adapted to gaming-specific needs.

• Treasure and Mage

Treasure, an Ethereum Layer 2 dedicated to Web3 gaming, recently launched Mage, its own launchpad for AI agents. Developed in partnership with ElizaOS, Mage was designed to facilitate the integration of autonomous agents within games built on Treasure. One of the first use cases of this technology was Smolworld, a next-generation Tamagotchi where small monkey AI agents evolve independently in a virtual environment.

Blockchains specialized in AI Gaming

Beyond frameworks, some blockchains are striving to establish themselves as reference infrastructures for games integrating AI agents. These layer 1 and layer 2 networks provide optimized environments to accommodate complex interactions between players and autonomous agents.

• Treasure + Beam

Treasure, already mentioned for its role in incubating AI agents, relies on the ZkSync technical stack, allowing it to efficiently process in-game interactions. Another major player is Beam, a Layer 1 operating as an Avalanche subnet, now recognized as one of the leading infrastructures for Web3 gaming. In December 2024, Beam announced the launch of BeamAI, a division dedicated to synergies between AI and gaming.

• Gameswift

Gameswift is a layer 2 specialized in AI with the launch of GS Force AI, a solution leveraging gamers' unused GPUs to power AI models. The project has also expanded its partnerships with Fetch AI, Aethir, and IO.net to optimize its AI agents' performance. More recently, Gameswift incubated the Game Dev Agent project, a platform designed to facilitate AI-assisted video game creation.

• Other emerging players

We can mention Ancient8, which partnered with Oraichain to integrate AI agents into its gaming Layer 2, and Sovrun, another Layer 2 on Hyperliquid, displaying a growing willingness to adopt AI agent technologies in its solutions.

Infrastructure and SDKs

In addition to dedicated blockchains, some of the projects mentioned above are also developing SDKs (Software Development Kits) that facilitate the integration of AI agents into Web3 games. It is worth noting that, in a way, frameworks are considered SDKs, meaning this category has somewhat blurred boundaries.

• NRN Agent

NRN Agent offers an AI Gaming SDK that enables designing, training, and deploying AI agents in video games. The focus is on imitation learning, allowing these agents to replicate human behaviors, and reinforcement learning, which helps them optimize their actions through trial and error. Since this process requires data, NRN Agent has developed its own data collection system.

In January, NRN Agent announced a partnership with Ronin, enabling the integration of AI agents into this blockchain's ecosystem. Finally, they have also teamed up with ai16z to explore potential synergies between their SDK and the ElizaOS framework.

• Zentry

Zentry aims to build the "metagame," an infrastructure interconnecting multiple games and players. Although this vision remains somewhat speculative for now, the project's architecture exists and is structured around three layers:

• zApp, an application layer.
• zAI, an infrastructure for coordinating AI agents.
• zProtocol, a protocol for interoperability and data management.

The project has started unveiling its first products, such as Nexus, an application linking humans and AI agents, and Radiant, a platform rewarding players in exchange for their gaming data.

Agentic Metaverse

As its name suggests, this category refers to projects developing virtual environments (such as metaverses) or tools for these environments, leveraging AI agent technologies.

• Hyperfy

Hyperfy is a platform that allows developers to easily create virtual worlds accessible to users via a browser or VR headset. The idea is to enable the creation of video games, bring communities together, or design experiences for specific events.

Additionally, Hyperfy incorporates AI agents capable of fulfilling various roles, from simple guides to interactive companions within metaverses. Like many projects mentioned here, this is still in the experimental stage, but several initiatives are already highlighted on their website.

• Nifty Island

Nifty Island allows users to create and explore virtual and customizable islands. The idea is to let users build virtual experiences or video games, such as races or battle royale-style matches, with access restricted to token or NFT holders. The project partnered with NRN Agent and ai16z to enable players to import autonomous AI agents into their environments.

• Other projects

We can mention Youmio, which launched Youmio Agent, a launchpad dedicated to 3D AI agents specifically designed to operate in virtual worlds. Additionally, Hytopia, a gaming platform that enables the creation of virtual worlds similar to Minecraft, provides an SDK that allows users to deploy AI agents into these metaverses quickly.

Gaming Agents

Gaming agents closely resemble the concepts discussed in the previous section. These are either AI agents capable of playing video games as human replacements or NPCs within video games.

• AiVeronica

AiVeronica is an AI agent aiming to become a true gaming influencer, capable of playing games during live streams, devising strategies, and interacting with a community in real time.

• GAMES.GG

GAMES.GG is developing an infrastructure to integrate AI into its Web3 gaming platform. They launched their AI agent, GAMER, designed to coach players, act as NPCs in certain games, or even play autonomously.

• Blockrot

Blockrot is an AI gaming agent specialized in Minecraft, capable of live streaming its gameplay and commenting on its actions in real-time. Its replays are available on Twitter.

Although these projects are still experimental, they foreshadow the emergence of fully integrated AI agents capable of participating in Web3 gaming mechanics.

AI Games

Some Web3 studios are beginning to integrate AI agents directly into gameplay, offering more dynamic and immersive gaming experiences.

• Colony (Parallel)

Parallel is a pioneer in AI gaming on Solana, as they have been discussing the idea of integrating autonomous agents for two years. The project is developing Colony, a survival simulation game where avatars (AI agents) are entirely autonomous. Unlike traditional NPCs, these entities can make their own decisions and evolve based on player actions.

• Other projects

We can mention Smolworld (developed by the Treasure studio), a project already discussed that aims to become a modern version of Tamagotchi, where each AI agent is an evolving monkey that players can interact with.

Additionally, Avarik Saga is working on Red Covenant, a mobile RPG featuring AI-powered NPCs with enhanced interactions. Notably, this studio is backed by Virtuals, the platform facilitating the easy launch and tokenization of AI agents.

Finally, Illuvium, a well-known Web3 game, recently signed a partnership with Virtuals to integrate AI agent NPCs to enrich the gaming experience.