Finly

Finly lets you build your own investment team, where agents like a researcher, analyst, trader, and advisor deliver personalized insights, helping beginners learn and invest confidently in less time.

Comment

Inspiration

When it comes to AI in stock investing, most solutions focus on experienced traders—like quantitative models and algorithmic bots. But there’s a growing overlooked group: busy professionals who want to learn investing on their own.

Existing solutions falls short in a few ways:

  • Misaligned incentives: Traditional fund managers may recommend assets that generate commissions rather than maximize investor value.
  • Lack of personalization: Robo-advisors often suggest generic strategies (e.g., ETFs) without considering individual goals, risk tolerance, or experience level.
  • Lack of transparency: Many tools provide recommendations or performance updates, but fail to explain the reasoning behind decisions. We believe beginner investors do not just need answers. They also want to understand "why" , with understandable reasoning they can learn from.

Target Users

  • Office workers and young professionals (22-40)
  • Beginner investors with basic financial knowledge

These users see investing as a way to diversify their savings and grow the unused portion of their monthly income. They tend to be long-term, fundamentals-driven investors who consistently allocate part of their salary into investments.

Their Needs

  • A simple way to learn investing while actively managing a portfolio
  • Guidance that is beginner-friendly and personalized to risk, goals, and portfolio

Pain Points

"As a full-time office worker, I want to learn investing on my own, but I lack the time and direction to start".

⇒ This is challenging because investing demands multiple cognitive skills—analysis, research, and critical thinking.

Introducing Finly

Finly is an AI-powered platform that lets you build your own investment team to receive personalized research, analysis, and actionable insights. Instead of relying on a single model, Finly uses multiple specialized agents that collaborate to support your investing journey:

Core Agent Roles

  • Market Researcher
    Curates macro trends and company news into concise, relevant insights to keep you up to date.

  • Analyst
    Interprets market information and evaluates whether an asset is bullish or bearish based on fundamental signals.

  • Technical Trader
    Identifies optimal entry and exit points using technical indicators and price signals.

  • Portfolio Manager
    Synthesizes insights from all agents and aligns them with the user’s portfolio, goals, and risk profile to make final recommendations.

Current product flow :

  • Onboarding + profile/risk capture
  • Conversational intake to clarify goals
  • Report generation per ticker
  • Multi-agent panel chat (streaming responses)
  • Portfolio view with sortable holdings and board-linked insights
  • Watchlist generated from board activity
  • Heartbeat Monitoring (Real-time price alerts with customizable triggers and automated actions)

How We Built It

Architecture

  • Mobile app: apps/mobile (Expo + React Native + NativeWind)
  • Backend API: apps/backend (FastAPI + SQLite + proxy/orchestration)
  • Agent runtime: apps/agents (TradingAgents-style multi-agent server)

Key Implementation Details

  • Streaming UX for intake and panel chat via SSE endpoints:
    • /api/intake/stream
    • /api/report/chat/stream
  • Persistent user/report/chat state in backend database
  • Market data endpoints for US ticker flows
  • Mobile board threads mapped into portfolio and watchlist screens

Challenges We Ran Into

  • Coordinating three runtimes (mobile, backend, agent server) within tight hackathon constraints
  • Handling degraded states cleanly when agent services or model providers are unavailable
  • Maintaining responsiveness while supporting long-running agent/report pipelines
  • Balancing product polish with architecture discipline under tight deadline constraints

Our Accomplishments

  • A fully working end-to-end multi-agent flow from intake -> report -> panel discussion
  • A clean mobile experience across Home, Portfolio, Board, and Settings
  • Real-time Streaming chat/report experiences instead of blocking "wait and refresh" UX
  • Practical local architecture that can deploy as two services (backend + agent server)

What We Learned

  • In fintech, Reliability and transparency are as important as model quality
  • Users trust recommendations more when rationale is visible and role-based
  • Multi-agent outputs need strong product orchestration to feel coherent
  • Fast fallback paths and clear error states are essential for demo resilience

What's Next For Finly

  • Expand beginner education loops (guided quests, confidence milestones)
  • Stronger memory and personalization across sessions
  • Extend coverage to Vietnamese markets with richer risk simulations
  • Add production-grade authentication, observability, and compliance layers
  • Strengthen deployment and data infrastructure for scalability
  • Extend heartbeat system to support multi-condition logic

Built With

Share this project:

Updates