FinWise Agentverse

Inspiration

Managing personal finances can be overwhelming, especially with fragmented tools for budgeting, saving, investing, and expense tracking. We wanted to create an all-in-one intelligent finance assistant powered by autonomous AI agents—making financial planning easy, interactive, and personalized. The rise of multi-agent systems and LLMs inspired us to integrate LangChain and HuggingFace with Fetch.ai’s Agentverse framework.

What it does

FinWise Agentverse is a multi-agent finance application that allows users to:

Register securely with validation checks.

Track their income, expenses, EMIs, and savings goals.

Visualize spending patterns with interactive charts (daily, weekly, monthly, yearly).

Get AI-powered investment advice.

Ask general financial questions answered by a finance-specialized LLM.

It uses autonomous agents for transactions, budgeting, market analysis, and LLM-driven advice, all working together to support users' financial decisions.

How we built it

Backend: Built using FastAPI with SQLite for data storage.

Frontend: Streamlit for a user-friendly UI and dynamic charting.

Agents: Created with Fetch.ai’s uAgents framework.

user_agent for authentication.

transaction_agent for storing and managing expenses.

budget_analyzer_agent for summarizing spending.

investment_advisor_agent powered by Hugging Face + LangChain.

knowledge_agent for financial Q&A using Falcon-7B LLM.

Visualization: Matplotlib and Plotly for rendering pie and bar charts.

Security: Password and email validations, token-based access, and secure agent communication.

Challenges we ran into

Syncing multiple agents in a real-time flow.

Validating secure login while maintaining session integrity.

Latency and formatting when integrating with large LLMs.

Designing a charting UI that’s responsive and intuitive with Streamlit.

Accomplishments that we're proud of

Fully functional AI agent ecosystem tailored for finance.

Seamless integration of LLMs with autonomous agent logic.

Clean frontend with charts that help visualize expenses and savings.

Extensible architecture to plug into banking APIs or more advanced investment modules.

What we learned

How to use Fetch.ai’s Agentverse ecosystem for real-world problems.

Effective ways to integrate HuggingFace and LangChain for domain-specific tasks.

Frontend/backend synergy for finance dashboards.

Handling authentication and role-based access in a decentralized agent setup.

What's next for FinWise Agentverse

Integrate real-time finance APIs like yFinance or Zerodha Kite.

Add support for crypto, mutual funds, and bank account sync.

Introduce notifications and agent alerts for budget breaches or savings goals.

Add a recommendation engine for insurance, loans, and investment products.

Port to mobile using Streamlit's experimental WebView or React Native.

Built With

• flask
• langchainpython
• streamlit
• uagent