Hivance

Inspiration

Retail investors are heavily influenced by global events like rate hikes, geopolitics, and tech breakthroughs, but rarely understand how these events translate into market movement. Most people end up reacting based on social media hype, which often leads to losses.

We wanted to build something that could help everyone understand the real impact of important events and get them thinking like an investor without needing a finance degree.

What it does

Hivance is an AI powered market analysis companion that lets users input any major world event and instantly receive:

Market effects
Portfolio specific impact for each stock they own
Clear recommendations like Buy, Hold, Sell, or DCA explained in simple English

Hivance deploys AI agents:

Stock Analyzer Agent
Geopolitical News Agent
Manager Agent

These agents each build their own report based on their expertise and the Manager Agent puts everything together and turns complex macroeconomic reasoning into insights.

How we built it

We designed a multi-agent system using an LLM orchestrator to break down the user’s event into structured reasoning steps:

Event analysis → classify the event and extract relevant economic dimensions
Sector impact prediction → estimate directional effects across industries
Stock reasoning → infer how each ticker is influenced using economic priors
Recommendation engine → synthesize findings into a simple action suggestion

We built a clean UI where users input an event and list their portfolio tickers. The backend then coordinates the agents, merges their outputs, and formats the final analysis.

Challenges we ran into

Divergent agent outputs: Getting multiple AI agents to cooperate without producing conflicting reasoning took several design iterations.
Converting vague events into structured data: The app needed to interpret ambiguous phrases like “AI disruption accelerates.”

Accomplishments that we're proud of

Produced a user experience that looks and feels like a professional analysis workflow
Built a fully multi agent financial reasoning system in under 24 hours
Created clear, hedgefund style market summaries from ambiguous natural-language events
Delivered portfolio specific recommendations with justifications

What we learned

Handling multiple API calls
Designing clean UI
Using Langgraph to allow agents to create complex financial logic
Integrating code from multiple people

What’s next for Hivance

We see Hivance evolving into a more complex investing assistant with more specialized agents:

Full portfolio syncing
Event alerts (“Oil prices spike → here’s how your energy holdings are affected”)
Sentiment analysis from social media and news pipelines

Our long term vision:
Provide everyone with the knowledge and confidence of a professional market analyst without needing to guess.