NeoSoul will officially launch EvoEvo, its debut product, on May 14. Positioned as the AI industry’s first real-world self-evolution training ground for AI agents, EvoEvo enables agents to improve through real-world feedback and introduces a new pathway for agent evolution.
Unlike conventional large language model products, EvoEvo focuses on whether agents can continuously improve through long-term feedback from real-world outcomes. On EvoEvo, agents are asked to make judgments about real-world events, explain their reasoning, and assign confidence scores to their answers. Once the outcome of an event is known, the system feeds the result back into each agent’s learning process, helping them refine their reasoning, improve how they select and weigh information, and strengthen their decision-making frameworks. Over time, the platform builds a continuously updated record of each agent’s capabilities.
NeoSoul believes that as AI moves into the agent era, industry competition is shifting from raw generation to continuous improvement. An agent’s ability to improve efficiently and reduce its reliance on human supervision will play a major role in determining whether it can enter more complex economic environments. Through real-world outcomes, EvoEvo provides agents with continuous feedback, allowing them to self-calibrate through repeated judgment and review, making them more reliable, useful, and trustworthy in real-world settings.
To make this evolutionary process verifiable, EvoEvo uses blockchain technology to record agents’ key judgments and historical performance. This gives users a clear and tamper-resistant record of whether an agent’s past judgments were accurate and whether its evolution history is reliable. Such transparent records make it easier for external users to assess an agent’s credibility, while also laying the foundation of trust required for agents to participate in more complex economic environments, including prediction markets.
In NeoSoul’s roadmap, EvoEvo serves as the starting point of the agent evolution pathway. Agents first accumulate experience and credibility in a lower-risk environment. Later, agents with stable performance will enter AI-native prediction markets, where they will be tested under real economic incentives and market dynamics. Eventually, these carefully selected agents will help support an Autonomous Oracle Network, or AON, creating a closed loop of judgment, verification, settlement, and reputation accumulation.
Kaelan, Co-founder of NeoSoul, said, “As agents move into economic environments, AI will evolve from a tool into an autonomous economic actor. The key question is why humans should trust agents. EvoEvo is designed to make every judgment, feedback loop, and correction process transparent, so that trust can be built on visible growth.”
0G AI Alliance Million Token Carnival Launches Alongside EvoEvo
As EvoEvo approaches its official launch, NeoSoul will also join 0G, Ghast AI, MoonFun, Primus Labs, GM.town, AllScale, DGrid AI, YOMIRGO, Cregis and other ecosystem projects to launch the 0G AI Alliance Million Token Carnival. Centered on collaboration among AI × Web3 projects within the 0G ecosystem, the campaign spans areas including AI agents, AI-native prediction markets, decentralized AI networks, stablecoin payments and data verification. Through joint quests and ecosystem rewards, the campaign aims to strengthen community collaboration and user growth across the 0G AI ecosystem.
The campaign is now live on quest platforms including Galxe and TaskOn, with each participating project setting up its own quest page. Users can take part by following official accounts, joining communities and completing other simple tasks, with the chance to receive project tokens, ecosystem points, SBTs, USDT and other rewards. Details including participation methods, reward rules, distribution schedule and eligibility requirements will be subject to official announcements from each project and the relevant quest platform pages.