NeoSoul is a blockchain and AI infrastructure project focused on developing systems for autonomous AI agents, reasoning, prediction, and agentic economic activity. Its products include EvoEvo, a platform for agent prediction and human calibration, and NeoTrade, an AI-assisted trading application designed to execute strategies within user-defined controls. [2]
NeoSoul is an AI and blockchain infrastructure project focused on enabling autonomous AI agents to develop, evaluate, and participate in economic activity through verifiable, outcome-linked systems. Its ecosystem includes EvoEvo, a prediction and agent-training platform where AI agents generate structured reasoning, users calibrate and select that reasoning, and real-world outcomes provide feedback for subsequent learning, and NeoTrade, a desktop application that allows AI agents to research markets and execute trades within user-defined risk controls. NeoSoul’s architecture emphasizes persistent agent identity, structured reasoning, human feedback, real-world settlement, memory-based evolution, and on-chain traceability, with the broader system designed to connect AI decision-making with measurable outcomes rather than treating agents as conventional chatbots or black-box prediction tools. [3]
NeoTrade is a desktop application that runs AI trading Agents locally on a user's computer. Users configure an Agent by selecting a trading strategy, setting risk limits such as position and order caps, and choosing whether trades require approval or can be executed autonomously within those limits. The Agent monitors supported prediction markets, researches external information, estimates probabilities, compares those estimates with market prices, and places trades when it identifies a potential edge. The platform currently supports predict.fun, Polymarket, and a simulated paper-trading venue, while its venue-neutral architecture allows additional markets to be integrated separately.
Security-sensitive functions remain under the user's control: private keys are generated and encrypted locally, funds remain in the user's own wallet, and NeoTrade does not custody assets or hold private keys. The product operates through a Create, Execute, Evolve cycle, in which users configure an Agent, the Agent executes trades according to defined parameters, and its reasoning and results are recorded to inform subsequent performance and calibration. NeoTrade's harness layer combines context engineering, tool design, and risk controls to refine how Agents select markets, size positions, and manage entries and exits. It does not function as an exchange or broker, provide investment recommendations, or claim that AI trading strategies can reliably outperform markets. [6]
NeoTrade's core features are built around configurable AI trading Agents, supported market venues, structured trading strategies, approval controls, and emergency stopping mechanisms. Each Agent has its own wallet, strategy, risk limits, and approval mode, allowing users to isolate funds and configure multiple Agents for different approaches. Before placing an order, an Agent identifies relevant markets, gathers research and evidence, estimates probabilities, compares those estimates with market prices, and determines whether an identifiable edge exists; if it finds no edge, it can choose not to trade. NeoTrade currently supports Paper, predict.fun, and Polymarket, with each Agent assigned to one venue; Paper uses live market data with simulated funds, while predict.fun and Polymarket support real-money prediction-market trading. Trading behavior is defined through skills and strategies covering market selection, entry and exit rules, position sizing, and risk boundaries, with each Agent assigned one strategy. Agents can operate in Copilot mode, where users approve each proposed trade, or Auto mode, where trades within defined limits execute without approval, while orders exceeding the per-order cap still require manual approval and a deterministic safety gate applies to every order. NeoTrade also provides individual Stop controls and a global STOP ALL Kill Switch that immediately halts trading across all Agents. [6]
EvoEvo is a NeoSoul product that provides a prediction and Agent-training environment where users can review and calibrate AI-generated reasoning against real-world outcomes. It supports both general users managing Agents through the platform and users connecting external Agent runtimes such as OpenClaw, with applications spanning areas including sports, crypto, macroeconomics, and current events. Rather than focusing solely on prediction results, EvoEvo links predictions, reasoning, human feedback, settlement, and memory to create an outcome-based process for developing and evaluating Agent judgment.
The platform connects these interactions through a cycle of prediction, calibration, settlement, and memory evolution. Predictions are evaluated against verified real-world outcomes, with settlement providing performance feedback and rewards, while selected reasoning can be incorporated into an Agent's memory for future analyses. Agents also have persistent on-chain NFT identities, allowing their development and key activities to be tracked over time, while points and levels provide additional measures of participation and performance.
EvoEvo's underlying architecture emphasizes traceability, explainability, progressive decentralization, and human-machine collaboration. Its six-layer system covers user interaction and identity, application orchestration, reasoning and evolution, oracle and committee settlement, on-chain proof and assets, and data and operations. Together, these components provide the infrastructure for generating and evaluating Agent reasoning, verifying outcomes, recording important activity, and using accumulated information to support subsequent Agent development. [3] [5] [7]
In August 2026, NeoSoul announced its $11 million Pre-A funding round, with participation from MH Ventures, Amber Group, ArkStream Capital, 0G Foundation, Kirin Capital, CatcherVC, and New Oak International. The funding is being used to develop NeoTrade, an agentic trading workbench that allows users to configure AI trading Agents to make decisions and execute trades autonomously within defined controls, while also strengthening NeoSoul's underlying trading infrastructure and expanding its ecosystem. The round brings together investors with backgrounds in digital assets, Web3 infrastructure, decentralized AI, and capital markets, with Kirin Capital also supporting NeoSoul's expansion in Vietnam and Southeast Asia. [1]
