This project provides an open-source research assistant that automates the generation of AI-powered reports from web search results. It targets researchers, students, and professionals needing to synthesize information from multiple sources, offering flexibility in AI model selection and search providers. The key benefit is a streamlined, customizable workflow for deep research and report creation.

How It Works

The application follows a three-step process: 1) Search Results Retrieval using configurable Google Custom Search or Bing Search APIs. 2) Content Extraction from selected search results via JinaAI. 3) Report Generation using user-selected AI models (Gemini, GPT, Claude, DeepSeek, local models via Ollama) and custom prompts. This multi-AI platform approach allows users to leverage the best model for their specific needs, while JinaAI ensures efficient content processing.

Quick Start & Requirements

• Install: npm install (or yarn, pnpm, bun)
• Run: npm run dev (or yarn, pnpm, bun dev)
• Prerequisites: Node.js 20+, API keys for chosen AI platforms (Gemini, OpenAI, Anthropic, DeepSeek, OpenRouter) and search providers (Google Custom Search, Bing Search), Upstash Redis for rate limiting.
• Setup: Requires obtaining and configuring API keys and potentially setting up Redis.
• Docs: Demo Video (link placeholder, actual link not provided in README)

Highlighted Details

• Supports local file uploads (TXT, PDF, DOCX) for integrated research.
• Features a "Flow" mode for recursive research, visual mapping, and report consolidation.
• Integrates with local models via Ollama, bypassing external API rate limits.
• Customizable via lib/config.ts for search providers, AI models, rate limits, and prompts.

Maintenance & Community

• The project is maintained by btahir.
• Contributions via pull requests are welcome; issues should be opened for major changes.
• Twitter handle provided for following the author's projects.

Licensing & Compatibility

• License: MIT
• Compatibility: Permissive MIT license allows for commercial use and integration into closed-source projects.

Limitations & Caveats

• Requires significant setup with multiple API keys and potentially Redis.
• Advanced reasoning models may require increased serverless function duration limits depending on the hosting platform.
• Demo video is sped up for brevity, actual processing times may vary.