GitHub - Ricko12vPL/QuantBrief: AI-powered market intelligence agent — 5-model Mistral pipeline analyzes SEC filings (256K context), screens news, reasons about portfolio impact, and delivers scheduled audio briefings in 5 languages. Built for Mistral AI Hackathon 2026.

AI-Powered Real-Time Market Intelligence Agent

Your personal CIO — multi-agent pipeline that screens news, analyzes SEC filings with 256K context, reasons about portfolio impact, and delivers scheduled audio briefings in 5 languages.

Built for Mistral AI Worldwide Hackathon 2026 · Track: Anything Goes · Kacper Saks

⚠️ Temporary hackathon URL — may be taken down after judging.
To register, enter any email and password (min. 8 chars) — no real email required.

The Problem

150+ million retail investors face a massive information asymmetry. A Bloomberg Terminal costs $25,200/year. Without it:

300+ financial articles/hour — impossible to read manually
SEC filings in legal jargon — 80-120 page 10-Ks are unreadable for most
No cross-source synthesis — data exists everywhere but nobody connects the dots
Language barriers — European investors face English-only sources

The Solution

QuantBrief is a multi-agent AI system that acts as your personal Chief Investment Officer. It monitors markets, SEC filings, and news — then delivers actionable, synthesized intelligence briefs on a schedule you define.

You schedule a daily brief at 09:00 UTC. QuantBrief automatically:

  1. Screens overnight news using Ministral 3B (~50ms/article)
  2. Detects material events affecting your portfolio & watchlist
  3. Analyzes full SEC filings in Mistral Large 3's 256K context window
  4. Reasons about portfolio impact using Magistral chain-of-thought
  5. Generates an audio briefing via ElevenLabs in your language
  6. Pushes the brief to your dashboard — ready when you wake up

Key Features

Multi-Agent AI Pipeline (4 Mistral Models + ElevenLabs)

Agent	Model	Role
News Screener	Ministral 3B	High-throughput filtering of news articles (<50ms each)
Filing Analyst	Mistral Large 3 (256K)	Full SEC 10-K/10-Q analysis in a single pass — no chunking, no RAG
Reasoning Engine	Magistral Medium	Chain-of-thought portfolio impact assessment with confidence scores
Synthesizer	Mistral Large 3	Cross-source correlation into executive brief + action items
Voice Agent	ElevenLabs	Professional multilingual audio briefings (EN/FR/DE/PL/ES)

Scheduled Brief Generation

Automatic recurring analysis — no manual clicking required:

Ticker sources: Custom tickers, Portfolio holdings, Watchlist, or All combined
Frequencies: Every 4 hours, Daily, Weekly
Configurable time: Set exact UTC hour/minute for daily and weekly runs
Concurrency-safe: Shared asyncio lock prevents overlapping pipeline runs
Persistent: Schedules survive server restarts (PostgreSQL-backed)

Earnings Call Analysis

Upload earnings call audio files — Mistral multimodal transcribes (Voxtral-ready via env config), Mistral Large 3 extracts financial highlights, forward guidance, risk factors, and Q&A insights.

Real-Time Market Data

Live quotes via yfinance — prices, change %, intraday data
Technical analysis — RSI, MACD, Bollinger Bands, moving averages with AI interpretation
Interactive price charts — candlestick and line views with 1D/1W/1M/3M intervals
News sentiment — per-ticker news with AI-scored sentiment and relevance
Financial ratios — P/E, P/B, ROE, D/E, profit margins with radar chart comparison

Multilingual (5 Languages)

Full UI and audio briefings in English, French, German, Polish, Spanish. Leverages Mistral's native multilingual capabilities to translate English-only SEC content.

Production-Ready Infrastructure

Deployed — Frontend on Vercel, Backend on Railway (PostgreSQL + Redis)
JWT Authentication — access/refresh tokens, bcrypt password hashing, secure login/register flow
PostgreSQL — persistent storage for users, briefs, portfolios, watchlists, schedules
Rate Limiting — per-endpoint limits (SlowAPI) to prevent abuse
Security Headers — CSP, X-Frame-Options, Referrer-Policy, Permissions-Policy
33 Automated Tests — auth, portfolio, watchlist, schedule isolation (IDOR prevention)
CI/CD — GitHub Actions for backend (pytest) + frontend (lint, TypeScript, build)

Architecture

DATA SOURCES                    AGENT PIPELINE                        OUTPUT

SEC EDGAR ──┐                  ┌─────────────────────┐               ┌──────────┐
yfinance ───┤  Ingestion  ──▶ │ Stage 1: SCREENING   │               │ React    │
RSS News ───┤  Layer          │ Ministral 3B (parallel)│──▶          │ Dashboard│
FRED ───────┘                  └─────────┬───────────┘    │          └──────────┘
                                         ▼                │          ┌──────────┐
                               ┌─────────────────────┐    │          │ Audio    │
                               │ Stage 2: ANALYSIS    │    ├────────▶│ Briefing │
                               │ Mistral Large 3 256K │    │          └──────────┘
                               └─────────┬───────────┘    │          ┌──────────┐
                                         ▼                │          │ JSON API │
                               ┌─────────────────────┐    │          │ REST+WS  │
                               │ Stage 3: REASONING   │────┘          └──────────┘
                               │ Magistral (CoT)      │
                               └─────────┬───────────┘
                                         ▼
                               ┌─────────────────────┐
                               │ Stage 4: SYNTHESIS   │
                               │ Mistral Large 3      │
                               └─────────┬───────────┘
                                         ▼
                               ┌─────────────────────┐
                               │ Stage 5: VOICE       │
                               │ ElevenLabs TTS       │
                               └─────────────────────┘

System Design

┌─────────────────────────────────────────────────────────┐
│                    Frontend (React 19)                   │
│  React Router · Zustand · Tailwind · Recharts · i18n    │
│  Auth Pages · Dashboard · Analysis Panels · Audio Player│
└──────────────────────┬──────────────────────────────────┘
                       │ REST API + WebSocket
┌──────────────────────▼──────────────────────────────────┐
│                  FastAPI Backend                         │
│  ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌───────────┐  │
│  │ JWT Auth │ │ Rate     │ │ Security │ │ CORS      │  │
│  │ Middleware│ │ Limiter  │ │ Headers  │ │ Origins   │  │
│  └──────────┘ └──────────┘ └──────────┘ └───────────┘  │
│  ┌──────────────────────────────────────────────────┐   │
│  │          25+ API Endpoints (REST + WS)           │   │
│  │  Auth · Brief · Schedule · Portfolio · Watchlist  │   │
│  │  Market · Filing · Audio · WebSocket Progress     │   │
│  └──────────────────────────────────────────────────┘   │
│  ┌──────────────────────────────────────────────────┐   │
│  │     5-Agent Pipeline (Orchestrator + Lock)        │   │
│  │  Screener → Analyst → Reasoner → Synth → Voice   │   │
│  └──────────────────────────────────────────────────┘   │
│  ┌──────────────────────────────────────────────────┐   │
│  │     APScheduler (Cron/Interval Triggers)          │   │
│  └──────────────────────────────────────────────────┘   │
└──────┬──────────────┬──────────────┬────────────────────┘
       │              │              │
  ┌────▼────┐   ┌─────▼─────┐  ┌────▼────┐
  │ Postgres│   │   Redis    │  │  W&B    │
  │ 6 tables│   │   Cache    │  │ Logging │
  └─────────┘   └───────────┘  └─────────┘

Tech Stack

Backend

Component	Technology
Language	Python 3.12
Framework	FastAPI 0.115 + Uvicorn (async)
Database	PostgreSQL 16 + SQLAlchemy 2.0 (async ORM)
Migrations	Alembic
Auth	JWT (PyJWT) + bcrypt
AI Models	mistralai SDK (Ministral 3B, Mistral Large 3, Magistral Medium)
Voice	ElevenLabs SDK (multilingual TTS)
Scheduler	APScheduler (AsyncIOScheduler, in-process)
Market Data	yfinance (quotes, candles, financials, news)
Cache	Redis 7 (with in-memory fallback)
Rate Limiting	SlowAPI (60 req/min global + per-endpoint)
Security	CSP, X-Frame-Options, Referrer-Policy, input validation
Tracking	Weights & Biases
Testing	pytest 8.3 + pytest-asyncio (33 tests)

Frontend

Component	Technology
Framework	React 19 + TypeScript 5.9
Build	Vite 7.3 (lazy-loaded route code splitting)
Routing	React Router 7.5
Styling	Tailwind CSS 4.2 (glassmorphism dark theme)
State	Zustand 5.0 (5 stores)
Charts	Recharts 3.7 + Lightweight Charts 4.2
i18n	react-i18next (EN/FR/DE/PL/ES — 150+ keys)
Real-time	WebSocket (pipeline progress)
Icons	Lucide React

Data Sources

Source	Data	Cost
SEC EDGAR	10-K, 10-Q, 8-K filings	Free
yfinance	Quotes, candles, financials, news	Free
Finnhub	Ticker search, company profiles	Free tier
FRED	GDP, CPI, rates, macro indicators	Free
RSS	Reuters, MarketWatch headlines	Free

Quick Start

Prerequisites

Python 3.12+, Node.js 20+, PostgreSQL 16+, Redis
API keys: Mistral AI, ElevenLabs (free tiers available)

1. Clone & Install

git clone https://github.com/Ricko12vPL/QuantBrief.git
cd QuantBrief

# Backend
cd backend
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

# Frontend
cd ../frontend
npm install

2. Configure

cp .env.example backend/.env
# Edit backend/.env — set at minimum:
#   MISTRAL_API_KEY=your_key
#   ELEVENLABS_API_KEY=your_key
#   JWT_SECRET_KEY=$(openssl rand -hex 32)
#   DATABASE_URL=postgresql+asyncpg://quantbrief:quantbrief@localhost:5432/quantbrief

3. Database Setup

# Create PostgreSQL role and database
psql -U postgres -c "CREATE ROLE quantbrief WITH LOGIN PASSWORD 'quantbrief';"
psql -U postgres -c "CREATE DATABASE quantbrief OWNER quantbrief;"

# Run migrations
cd backend
alembic upgrade head

4. Run

# Terminal 1: Backend
cd backend
uvicorn app.main:app --reload --port 8000

# Terminal 2: Frontend
cd frontend
npm run dev

5. Open

Navigate to http://localhost:5173 — register an account and start generating briefs.

Docker Compose (Alternative)

cp .env.example .env
# Edit .env with your API keys
docker compose up
# Frontend: http://localhost:5173
# Backend API: http://localhost:8000

Project Structure

QuantBrief/
├── backend/
│   ├── app/
│   │   ├── main.py                     # FastAPI entry + lifespan + middleware stack
│   │   ├── config.py                   # Pydantic settings (env-driven)
│   │   │
│   │   ├── agents/                     # 5-agent AI pipeline
│   │   │   ├── orchestrator.py         # Pipeline coordinator with progress callbacks
│   │   │   ├── news_screener.py        # Ministral 3B — parallel news filtering
│   │   │   ├── filing_analyst.py       # Mistral Large 3 — 256K SEC analysis
│   │   │   ├── reasoning_engine.py     # Magistral — chain-of-thought reasoning
│   │   │   ├── synthesizer.py          # Mistral Large 3 — brief generation
│   │   │   ├── voice_agent.py          # ElevenLabs TTS
│   │   │   └── earnings_transcriber.py # Multimodal earnings call analysis
│   │   │
│   │   ├── api/                        # 25+ REST + WebSocket endpoints
│   │   │   ├── routes_auth.py          # Register, login, refresh, me
│   │   │   ├── routes_brief.py         # Generate, latest, history
│   │   │   ├── routes_schedule.py      # CRUD + pause/resume
│   │   │   ├── routes_portfolio.py     # Portfolio management
│   │   │   ├── routes_watchlist.py     # Watchlist + ticker search
│   │   │   ├── routes_market.py        # Quotes, technicals, ratios, news
│   │   │   ├── routes_filing.py        # SEC filing analysis + earnings calls
│   │   │   ├── routes_audio.py         # Audio generation
│   │   │   └── ws_realtime.py          # WebSocket pipeline progress
│   │   │
│   │   ├── auth/                       # Authentication layer
│   │   │   ├── jwt.py                  # PyJWT access/refresh token management
│   │   │   ├── password.py             # bcrypt hashing
│   │   │   ├── dependencies.py         # FastAPI Depends (get_current_user)
│   │   │   ├── schemas.py              # Register/Login/Token Pydantic models
│   │   │   └── usage.py                # Usage tracking (future metering)
│   │   │
│   │   ├── db/                         # Database layer
│   │   │   ├── engine.py               # Async engine + session factory
│   │   │   └── models.py              # 6 SQLAlchemy ORM tables
│   │   │
│   │   ├── services/
│   │   │   └── scheduler.py            # APScheduler (CRUD, execution, DB persistence)
│   │   │
│   │   ├── models/                     # Pydantic data schemas
│   │   ├── data_sources/               # SEC EDGAR, yfinance, Finnhub, FRED, RSS
│   │   ├── analytics/                  # Technical signals, ratios, sentiment
│   │   ├── middleware/                 # Security headers middleware
│   │   └── utils/                      # Cache, rate limiter, validation, W&B logger
│   │
│   ├── prompts/                        # Versioned Mistral prompts (7 .md files)
│   ├── alembic/                        # Database migrations
│   ├── tests/                          # 33 pytest tests (4 test files)
│   └── requirements.txt                # 24 Python dependencies
│
├── frontend/
│   ├── src/
│   │   ├── App.tsx                     # React Router (lazy-loaded routes)
│   │   ├── components/
│   │   │   ├── Auth/                   # LoginPage, RegisterPage, ProtectedRoute
│   │   │   ├── Landing/               # Hero page
│   │   │   ├── Dashboard/             # 7 components (BriefCard, PriceChart, etc.)
│   │   │   ├── Analysis/             # 7 components (Sentiment, Technicals, Filing, etc.)
│   │   │   ├── Audio/                # AudioPlayer, TranscriptView
│   │   │   └── Common/              # WatchlistManager, ScheduleManager, etc.
│   │   ├── stores/                   # 5 Zustand stores (auth, brief, portfolio, etc.)
│   │   ├── lib/                      # Fully typed API client + WebSocket
│   │   └── i18n/                     # 5 locale JSON files
│   └── package.json
│
├── .github/workflows/ci.yml           # CI: pytest + lint + tsc + build
├── docker-compose.yml                  # PostgreSQL + Redis + Backend + Frontend
└── .env.example

Database Schema (6 Tables)

users                  -- id, email, hashed_password, display_name, tier, created_at
briefs                 -- id, user_id, data (JSON), sentiment, audio_url, generated_at
portfolio_positions    -- id, user_id, ticker, shares, avg_price (unique per user+ticker)
watchlist_items        -- id, user_id, ticker, company_name (unique per user+ticker)
schedules              -- id, user_id, frequency, hour, tickers, paused, next_run_at
usage_logs             -- id, user_id, action, metadata, created_at (future metering)

API Reference

Authentication

Method	Endpoint	Auth	Description
POST	`/api/auth/register`	No	Create account, returns JWT tokens
POST	`/api/auth/login`	No	Login, returns JWT tokens
POST	`/api/auth/refresh`	No	Refresh access token
GET	`/api/auth/me`	Yes	Current user info

Brief Generation

Method	Endpoint	Auth	Description
POST	`/api/brief/generate`	Yes	Run the full 5-agent pipeline
GET	`/api/brief/latest`	Yes	Most recent brief
GET	`/api/brief/history`	Yes	Paginated brief history

Scheduling

Method	Endpoint	Auth	Description
GET	`/api/schedule`	Yes	List user's schedules
POST	`/api/schedule`	Yes	Create schedule (max 10 per user)
PATCH	`/api/schedule/{id}`	Yes	Update schedule
DELETE	`/api/schedule/{id}`	Yes	Delete schedule
POST	`/api/schedule/{id}/pause`	Yes	Pause schedule
POST	`/api/schedule/{id}/resume`	Yes	Resume schedule

Market Data

Method	Endpoint	Auth	Description
GET	`/api/market/quotes?tickers=AAPL,NVDA`	No	Live quotes
GET	`/api/market/{ticker}/candles`	No	OHLCV candles
GET	`/api/market/{ticker}/technical`	No	Technical indicators
GET	`/api/market/{ticker}/ratios`	No	Financial ratios
GET	`/api/market/{ticker}/news`	No	News with AI sentiment
POST	`/api/market/{ticker}/analyze-technical`	Yes	AI technical analysis

Portfolio & Watchlist

Method	Endpoint	Auth	Description
GET/POST/DELETE	`/api/portfolio`	Yes	Portfolio CRUD
GET/POST/DELETE	`/api/watchlist`	Yes	Watchlist CRUD
GET	`/api/watchlist/search?q=...`	No	Ticker search

Filing & Audio

Method	Endpoint	Auth	Description
GET	`/api/filing/{ticker}`	No	List recent filings
GET	`/api/filing/{ticker}/analyze`	Yes	AI filing analysis
POST	`/api/filing/earnings-call`	Yes	Upload & analyze earnings call
POST	`/api/audio/generate`	Yes	Generate audio briefing

Why This Wins

Full Mistral Ecosystem (4 Models)

Not a single-API-call project. Each model chosen for specific strengths:

Ministral 3B for speed — screening at ~50ms/article
Mistral Large 3 for depth — entire 10-K in 256K context, no chunking
Magistral for reasoning — transparent chain-of-thought with confidence scores
Mistral Large 3 for synthesis — cross-source correlation into actionable briefs

256K Context = No RAG Needed

Most tools truncate or chunk SEC filings. QuantBrief loads the entire 80-120 page filing into a single Mistral Large 3 call. Full context = better analysis.

Real Data, Real Impact

No synthetic demos. Live market data from yfinance, real SEC filings from EDGAR, actual news sentiment. The briefs analyze your actual portfolio in real time.

Scheduled Automation

Not just on-demand — fully automated brief generation on a schedule. Set it and forget it. Your morning brief is ready before you wake up.

Production-Grade Engineering

Not a prototype — production infrastructure with JWT auth, PostgreSQL persistence, rate limiting, security headers, 33 automated tests, and CI/CD. Ready to deploy.

Codebase Metrics

Metric	Value
Total lines of code	~13,500
Backend Python files	67
Frontend TypeScript files	36
React components	20+
API endpoints	25+
Database tables	6
Automated tests	33
Languages supported	5 (EN/FR/DE/PL/ES)
AI models used	4 Mistral + 1 ElevenLabs
Versioned prompts	7

Sponsor Prize Alignment

Prize	Qualification
Jump Trading (Quant Finance)	Core domain — financial analysis with real market data, technical indicators, SEC filings
ElevenLabs ($2K credits)	Audio briefing as first-class feature — multilingual TTS integrated into the pipeline
Best Agent Skills	5-agent pipeline: parallel screening + sequential deep analysis + reasoning + synthesis + voice
W&B	Full experiment tracking of agent decisions, token usage, and pipeline latency metrics

Hackathon Submission

Field	Value
Hackathon	Mistral AI Worldwide Hackathon 2026
Track	Anything Goes
Team	Kacper Saks
Company	KS Venture Group
Location	Online (Warsaw, Poland)
Mistral Models	Mistral Large 3 (256K), Magistral Medium, Ministral 3B
Sponsor Tech	ElevenLabs (Voice AI), W&B (Experiment Tracking)
Data Sources	SEC EDGAR, yfinance, Finnhub, FRED, RSS
Repo	github.com/Ricko12vPL/QuantBrief
Live Demo	frontend-omega-seven-11mo2mlvze.vercel.app (temporary)
Deployment	Frontend: Vercel, Backend: Railway (PostgreSQL + Redis)

License

MIT License

Democratizing financial intelligence, one morning brief at a time.

Name		Name	Last commit message	Last commit date
Latest commit History 22 Commits
.github/workflows		.github/workflows
backend		backend
demo-video		demo-video
frontend		frontend
.env.example		.env.example
.gitignore		.gitignore
README.md		README.md
docker-compose.yml		docker-compose.yml
package.json		package.json
screenshot.png		screenshot.png

Ricko12vPL/QuantBrief

Folders and files

Latest commit

History

Repository files navigation

AI-Powered Real-Time Market Intelligence Agent

The Problem

The Solution

Key Features

Multi-Agent AI Pipeline (4 Mistral Models + ElevenLabs)

Scheduled Brief Generation

Earnings Call Analysis

Real-Time Market Data

Multilingual (5 Languages)

Production-Ready Infrastructure

Architecture

System Design

Tech Stack

Backend

Frontend

Data Sources

Quick Start

Prerequisites

1. Clone & Install

2. Configure

3. Database Setup

4. Run

5. Open

Docker Compose (Alternative)

Project Structure

Database Schema (6 Tables)

API Reference

Authentication

Brief Generation

Scheduling

Market Data

Portfolio & Watchlist

Filing & Audio

Why This Wins

Full Mistral Ecosystem (4 Models)

256K Context = No RAG Needed

Real Data, Real Impact

Scheduled Automation

Production-Grade Engineering

Codebase Metrics

Sponsor Prize Alignment

Hackathon Submission

License

About

Topics

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages