AI Research Assistant

Live at: airesearchassistant.fly.dev

An AI-powered research assistant that conducts web searches, synthesizes findings using LLMs, and generates comprehensive multi-language reports.

Features

• Web Research: Automated web search via Tavily API with source analysis
• Multi-LLM Support: OpenAI, Anthropic, and DeepSeek with automatic fallback chains
• Streaming: LLM responses stream into live preview during research
• Multi-Language Reports: Research in English, translated to 12+ languages
• Semantic Cache: sqlite-vec powered vector search — similar queries hit cache automatically
• BYOK: Bring your own API keys via Settings page (never stored in database)
• CLI: Full command-line interface alongside the web UI
• Rate Limiting: Built-in rate limiting, request queuing, and graceful degradation
• Research History: SQLite-backed research tracking and analytics

Quick Start

1. Install

pip install -e .

2. Configure

cp .env.example .env
# Edit .env with your API keys (at least one LLM provider + Tavily)

3. Run

# Web UI
python app.py

# CLI
python cli.py "quantum computing trends 2025"
python cli.py "AI safety" --depth advanced --lang sl --output report.md

CLI Usage

python cli.py "research topic"              # Basic research
python cli.py "topic" --depth advanced      # Thorough search (2x credits)
python cli.py "topic" --lang sl             # Translate to Slovenian
python cli.py "topic" --focus "area1,area2" # Focus on specific areas
python cli.py "topic" --output report.md    # Save to file
python cli.py "topic" --no-cache            # Skip cache, force fresh

Project Structure

├── app.py                          # Flask web app (thin HTTP layer)
├── cli.py                          # CLI interface
├── src/
│   ├── services/                   # Business logic layer
│   │   └── research_service.py     # Core research orchestration
│   ├── agents/                     # Agent implementations
│   │   ├── research_agent.py       # Web research agent (streaming)
│   │   └── multilang_research_agent.py  # Multi-language research
│   ├── tools/                      # Tool integrations
│   │   ├── web_search.py           # Tavily web search
│   │   ├── search_cache.py         # Exact + semantic search cache
│   │   ├── embeddings.py           # Embedding providers (hash, OpenAI)
│   │   ├── report_writer.py        # Markdown report generation
│   │   └── translation.py          # Translation framework
│   ├── database/                   # Data persistence
│   │   ├── models.py               # SQLAlchemy models
│   │   └── sqlite_writer.py        # Research history writer
│   └── utils/                      # Utilities
│       ├── llm.py                  # LLM clients (OpenAI, Anthropic)
│       └── rate_limiting.py        # Rate limiting, fallback chains, streaming
├── templates/                      # Flask HTML templates
├── tests/                          # 156 tests (pytest)
└── docs/                           # Documentation & roadmap

API Keys

Service	Required	Purpose
Tavily	Yes	Web search
OpenAI / DeepSeek / Anthropic	At least one	LLM reasoning & streaming
Google Translate	No	Optional translation provider

Users can also bring their own keys via the /settings page (BYOK).

Development

# Install with dev dependencies
pip install -e ".[dev]"

# Run tests
pytest

# Format code
black . && isort .

Deployment

Deployed on Fly.io with GitHub Actions CI/CD. See docs/ROADMAP.md for full development history.