• @hyperfy_io just open-sourced its engines, enabling self-hostable multiplayer worlds with real-time building, drag-and-drop tools, interoperable assets, and more.

  

   

  • @virtuals_io shared it's partnership with @infinityg_ai  introducing multi-agent asset management, building the first AI-operated asset management company, introducing multi-agent asset management, building the first AI-operated asset management company, @Zenith_XBT, setting the ground for agent society (agent-to-agent interactions).

   

  

   

  • @virtuals_io announced a partnership with @avariksaga, enabling agents to livestream games, comment, and narrate in real-time. 

  
  

  • This follows their recent partnership with @illuviumio, which introduced sentient-NPCs/AI-NPCs into Illuvium games.

  

   

  • Beyond Virtuals ecosystem, we have also seen the rise of World State Machine that goes beyond traditional conversational agents by creating a persistent, 24/7 simulated environment where Kira exists as a digital persona with human-like behaviors.
  • Unlike typical AI, Kira doesn't just react to inputs; she lives within this environment, engaging in activities that mimic human life, such as studying, working, and socializing, based on a complex system that tracks time, weather, location, and interactions with other AI characters or NPCs like Mei, Keiko, and Izumi.

  

   

   

  The Reality of Westworld-Inspired Games

  While everyone talks about creating Westworld-like games (immersive worlds driven by dynamic NPCs), @powpowfun has been beta-testing its Virtual West World game for a while now. This game features AI agents powered by open-source models from FullmetalAI. Notably, Fullmetal recently integrated ElizOS, which has long-term memory enabled via RAG.

   

   

  The team plans to launch @indiedotfun in about a week. Indiedotfun is like @pumpdotfun but for games, allowing anyone to quickly build their own worlds using templates and the game/physics engine by @moddio. These worlds can then be tokenized, enabling users and communities to gain ownership of their favorite games.

   

  Introducing Indie: Tokenization of Agentic Metaverse(s)

  Indie is a platform that lets developers build and launch Web3-enabled worlds without needing crypto knowledge. It’s powered by the @moddio game engine, which already boasts a large Web2 community that builds thousands of games monthly.

   

  

  Creators can choose to launch a token, enabling backers to fund the game and receive tokens in return. These tokens act as the game’s in-game currency. Imagine Roblox, but instead of Robux, creators can use their own token as currency. Open-source LLMs from FullmetalAI are built-in, allowing creators to launch AI agents directly into the game, making the game a lot more dynamic and fun.

   

  Addressing Web3 Gaming Challenges

  One of the challenges in Web3 gaming is the lack of interest in AAA-level games. Web3 itself is a game—people find speculating on AI agent coins more fun than playing Web3 games.

  This creates a short attention span, with interest peaking for 2–3 days before moving on to the next thing.

   

  From the gaming studio side, there’s a significant risk: even if the game is fun, interest wanes if the token doesn’t pump. Conversely, if the token pumps but the game isn’t fun, people might engage briefly, but retaining attention remains difficult. Studios face a twofold challenge: delivering a compelling game while ensuring the token price goes up.

   

  Why Indie is Perfect for Web3 Attention Spans

   

  Indiedotfun is the perfect engine to cater to this short attention span. With

  @moddio, lightweight indie games can be shipped in days (instead of months), enabling creators to quickly deliver immersive worlds with dynamic AI-NPCs. Gamers can move from one indie game to another when bored, while creators improve their metaverse experiences to retain gamers and maintain the value of their game tokens.Imagine, instead of conversational agents yapping on Twitter, you’ll have Agentic Metaverse Worlds to play in, speculate on their tokens, and gain ownership of these worlds.

   

  

  $BNTY acts as the main currency powering everything in this ecosystem.

  Alpha Tease: @indiedotfun handles the supply side, while @ARCAgents handles the demand.

  (Hope the ARC team doesn’t kill me for teasing this lol.)

   

  Final Thoughts

  The Agentic Metaverse narrative is evolving rapidly. I expect more gaming studios and teams to leverage G.A.M.E and Eliza to power their games. While there are gaps in these frameworks that need addressing, unique gaming-focused solutions will likely emerge in the coming months.

   

  Additionally, I anticipate Reinforcement Learning and Vision Language Models to grow in prominence. The increasing B2B demand for dynamic AI in games positions @ARCAgents at the forefront with their Agent Gamification Engines.

   

  Disclaimer: This document is intended for informational & entertainment purposes only. The views expressed in this document are not, and should not be construed as, investment advice or recommendations. Recipients of this document should do their due diligence, taking into account their specific financial circumstances, investment objectives, and risk tolerance (which are not considered in this document) before investing. This document is not an offer, nor the solicitation of an offer, to buy or sell any of the assets mentioned herein.

   

  Link

  

2024 was the most exciting year yet for the Theta ecosystem, as we saw more adoption than ever of Theta among academia, the crypto community, and the corporate world. The key milestone of the past year was undoubtedly the release of Theta EdgeCloud, the first hybrid cloud computing platform built on a fully distributed edge architecture powered by Theta edge network. Launched in May 2024, EdgeCloud’s debut has been a smashing success, with nearly 20 marquee customers onboarding including the NHL’s Vegas Golden Knights, FlyQuest esports, University of Oregon, Seoul National University, Liner, JamCoding and more.

June 2024 saw the release of the Elite Booster upgrade (Edge Node v5.0.2) empowering users to earn even more TFUEL and partner TNT-20 tokens for performing GPU-intensive AI and video tasks. This release greatly increased edge node earnings and enabled all edge node operators to participate in next-gen AI, video and computing jobs on EdgeCloud.

These upgrades coincided with the strong growth in Theta edge network, with more than 20k daily active Edge Nodes across the world and over 30k monthly active nodes. We expect the edge network to continue to expand in 2025 in order to support the increasing number of EdgeCloud customers, and to that end in September 2024 we launched Theta Edge Node for Android devices running calibration jobs with mobile optimized AI object detection algorithms. Over the course of the year, the daily Metachain transactions on Theta network also increased by over 50% and is now approaching 350k transactions per day, driven primarily by our subchain partners Lavita, POGS, Passaways, Grove, Replay, Bullit and others.

In the next year, we envision Theta customers utilizing EdgeCloud, RAG chatbot and AI agents across academia, enterprises, professional sports, esports gaming to begin exploring and deploying their own custom subchains to support transaction intensive, blockchain enabled AI applications, GenAI text-to-video, text-to-3D, LLM chatbots and AI agents. This will be supported by the beta release of the Theta AI Agents platform in the first half of 2025.

Theta’s 2024 roadmap was significant, but we’re proud of the progress Theta made in achieving and even exceeding it. In H1, the EdgeCloud rollout was successful with cloud hosted nodes up and running, a suite of popular GenAI and LLMs, and a host of new partners onboarded in enterprise and academia. EdgeCloud also began offering Agentic AI as a service, empowering businesses with cutting-edge AI capabilities with its first Agentic AI product, the RAG (Retrieval Augmented Generation) Chatbot. Later in the year, persistent storage was added to meet EdgeCloud AI customer needs. The Elite+ Booster feature was rolled out for Elite Edge Nodes, and Studio-level DRM was added to Theta Video API. TDROP was also extended to use as part of OpenTheta marketplace. Though the vast majority of these milestones were hit on schedule, there were a few items on the roadmap that were delayed and pushed back for various reasons. For example, Theta Video API (TVA) was made part of the broader EdgeCloud platform, so the next major update to TVA will come as EdgeCloud itself fully rolls out in 2025. Support for EdgeCloud job prioritization will also come in a 2025 EdgeCloud update, as well as support for users running their own EdgeCloud nodes in data centers or their own personal cloud infrastructure. But aside from these revisions, the 2024 roadmap was largely achieved and we intend on the same for 2025.

Building on these successes, Theta development is pushing forward with an ambitious roadmap for 2025 and beyond. The team is now focused on the next main milestone, the beta release of the hybrid edge cloud architecture on June 25, 2025. This major release will support seamless job orchestration and optimization of compute-intensive workloads across cloud-based GPUs and Theta’s distributed edge nodes run by our community. The hybrid architecture will enable the platform to essentially run any computation jobs that can be “containerized” from AI, video encoding/transcoding, 3D rendering workloads, to GPU-intensive financial modeling, scientific drug research jobs and more.

The main challenge is that edge nodes are heterogeneous with differing specs, capabilities and geo-distributed with about ⅓ in North America, ⅓ in EMEA and the remainder in the rest of the world. These consumer grade GPUs including NVIDIA 3090s, 4090s and the newly announced 5000 RTX series have limited VRAM with various CPU configurations, and not always on which pose technical hurdles. The job orchestration and containerization layer must be intelligent and optimized for the target node with auto failover. The goal over time is to support AI inference including GenAI text-to-image, image-to-image, video transcoding, AI object detection, 3D rendering, chatbots and other AI agent workloads that can be spliced and completed in tandem.