WealthM Chatbot Main Page
WealthM Generates Widget
Politician Transaction
Generation Reason

WealthM

Inspiration

While there's an overwhelming amount of market data out there, 99% of it is just noise. Where's that crucial 1% of actionable insights you actually need to execute buy or sell trades?

We decided to deliver that 1%.

Unlike general-purpose LLMs like ChatGPT or Claude, WealthM provides accurate market data and generates insights tailored to the unique characteristics of financial data.

WealthM is your single, actionable market data agent for making buy or sell decisions – a revolutionary fintech service that's your investment JARVIS.

What it does

AI-Powered Watchlist: No more tedious ticker-by-ticker entry when setting up watchlists on traditional market data platforms. With WealthM, users can simply capture and drag-and-drop their favorite stocks, ETFs, or economic indicators to get started instantly.
Discovery Feature: Traditional market search and exploration feels intimidating for average users – it's like Yahoo Finance where you have to navigate between search and browse. This differs from the intuitive search-first UX we're used to with Google or chatbots. To solve this, we focused on US markets within the vast global market data landscape, leveraging Bolt.new and integrated third-party tools to develop this in just days.
Reliability: Chatbots often provide inaccurate data that can harm users – and in financial markets, this leads to catastrophic losses. WealthM is the world's first project to solve this through licensed data commercialization, ontology, specialized AI models, and proprietary MCP working seamlessly together.
Widget-Powered Analysis: When notable market events occur, investors often wonder "So what should I do?" WealthM goes beyond just reporting transactions – we perform AI clustering and multiple regression analysis on historical data to calculate price correlations. Users get actionable insights like "When this politician trades this type of stock, it historically moves X% in Y days on average."
Accurate News Summaries: Existing chatbots often miss crucial data or cite wrong sources when summarizing market news. WealthM uses securities market-specific ontology to generate metadata, understand relationships between data points, and accurately organize essential information.

All these features in one platform help investors spend less time searching for information and focus more on making better investment decisions.

WealthM supports the entire information consumption journey from discovery to search. By combining partnerships with Nasdaq Group, LSEG Data & Analytics, and Palantir for AI-ready market data, we've integrated Waker's MCP built on 6 years of ontologized global market databases with domain-specific AI through third-party integration to deliver reliable, actionable investment insights.

How we built it

WealthM leverages a cutting-edge tech stack to build a robust and scalable system. Our development process:

Frontend (FE)

Tailwind CSS: Used for managing UI styles. Tailwind's utility-first approach enabled rapid responsive design implementation.
Zustand: For state management, enabling efficient and simple state handling.
WebSocket: Handling real-time interactions with WAISOD LLM.
Vercel: Efficient deployment and management of our web application.
Lovable: Initially set up to convert Figma UI to code, but we rebuilt it from scratch after hitting limitations with LLM-provided codebase.

Backend (BE)

AWS: Cloud infrastructure for APIs and databases, ensuring reliable service delivery.
Java, Kotlin, Spring Boot: Core backend languages and frameworks for data processing and widget APIs.
PostgreSQL: Primary database for data storage.
Kafka: Queue and broker service for real-time price data and streaming, enabling real-time serving.
Docker: Container-based architecture for efficient deployment and management of APIs and workers.

AI Development

Redis: Communication between API server and WAISOD agents, serving as task and event queues.
- Task Queue: Distributes user tasks to agents
- Event Queue: Streams agent reasoning plans, tool results, and status to clients in real-time
WAISOD: Model for agent reasoning, processing through planning, reasoning, tool calling, and generation.
Llama-index: Event-driven workflow engine implementing various tools for efficient agent task execution.
MCP (Model Context Protocol): Enables agents to dynamically call and utilize tools.
GoogleSearch: Supporting market news and web-based analysis tools.

Challenges we ran into

Figma and Bolt.new Integration: When applying Figma designs to Bolt.new, we encountered broken designs and unexpected implementations. We initially built with Lovable, then rebuilt the codebase after hitting limitations.
LLM Service and Widget Integration: Technical challenges in integrating data visualization widgets with LLM services. Beyond technical issues, we also grappled with structuring responses for intuitive and efficient UI/UX.
AI Code Scalability: Recognizing that AI-generated code lacked scalability for future service expansion, we proactively worked to improve the codebase foundation.

Accomplishments that we're proud of

Actionable Market Insights: We focused on the reliability and utility of market Q&A that OpenAI, Anthropic, and Grok can't provide. By analyzing data relevant to user queries and processing it according to data types, we deliver trustworthy, tailored responses. This user-experience-focused insight delivery, initiated through Bolt.new, will become the foundation for the AI-era Yahoo Finance serving global investors.
Rapid Development and Deployment: Completed prototype development and deployment in just over a week. This demonstrates our team's augmented efficiency and rapid execution capabilities thanks to Bolt.new. Through continuous updates, we'll evolve this into a user service that delivers consistent WoW moments.

What we learned

Ease of Third-Party Tool Integration with Bolt.new: Integrating third-party tools in Bolt.new was surprisingly straightforward. This enabled smooth feature integration and efficient platform development.

What's next for WealthM

Using Bolt.new, we'll add user registration and login features to expand functionality and personalize user experience based on collected data. We'll also enhance our data sophistication for more refined insights. Bolt.new is the optimal platform for this work, playing a crucial role in feature expansion and data quality improvement.