As generative AI advances, autonomous agents are evolving from conversational chatbots into intelligent systems capable of aggregating enterprise data, executing workflows, and assisting high-stakes decision-making. In digital asset investment, market signals remain heavily fragmented across crypto exchanges, social networks, news outlets, and on-chain metrics. Investors spend significant effort cross-referencing information, while AI models lacking verifiable data sources and human oversight risk hallucinating or delivering misleading advice.
Addressing this trust deficit, team "sudo chmod 777" engineered "MAIA," a trusted AI investment assistant designed to distill complex market data into actionable, traceable insights. Competing in the "MaiCoin Group: Smart Wealth Management" category, the team won top honors at the 2026 Taiwan Generative AI Applications Hackathon, organized by DIGITIMES under the guidance of the Administration for Digital Industries (ADI), Ministry of Digital Affairs (MODA). Amazon Web Services (AWS) served as the AI technology provider, delivering the generative AI and cloud services that powered the solutions.
Integrating Diverse Data for Traceable AI Recommendations
"MAIA" integrates data from MaiCoin Group's MAX digital asset exchange, insights from Threads, market data, and on-chain information to build an AI agent that combines trusted data sources, social engagement, and analytical capabilities. Unlike conventional AI chatbots that rely primarily on large language models to answer questions, MAIA brings together internal business data, individual investment information, and external market data on a single platform. This enables the AI to go beyond simply answering questions by providing capabilities for analysis, reasoning, verification, and execution.
More specifically, the platform can connect to MAX exchange trading records, asset allocations, trading gains and losses, real-time market data, derivatives data, macroeconomic information, and Threads discussions to establish a comprehensive foundation for investment analysis. By linking each AI-generated recommendation to verifiable data sources, MAIA aims to make its recommendations traceable and enhance confidence in investment decision-making.
MAIA: Balancing Efficiency and Security with the "Harness" Governance Framework
In terms of trading strategies, MAIA also supports strategy backtesting and automated trading. Users can interact with the AI in natural language to develop investment strategies and conduct walk-forward backtesting. Once a strategy has been validated, it can be packaged into a trading bot, with the Runner connecting to real-time market data and automatically executing the trading workflow. Because strategy validation and live trading use the same signal sources, the approach not only significantly reduces the need for manual intervention but also enables AI to evolve from an analytical tool into an intelligent agent capable of executing tasks.
To prevent AI from generating inaccurate information or carrying out high-risk operations, the team developed a "Harness" governance framework that incorporates Prompt Routing, persona management, memory management, data validation, human approval, and response traceability. Each AI response can be accompanied by its data sources, reasoning, and applicable limitations, while high-risk operations require human approval before execution. This approach balances the efficiency of AI with the security and trustworthiness required for financial applications.
The team said that participating in an AWS workshop before the formal competition not only helped them quickly become familiar with relevant cloud services and AI development tools, but also accelerated the implementation of MAIA's full range of capabilities. The team's focus on building a "trusted AI agent" was a key factor in setting its entry apart from the competition.