KEA Research

KEA Research

Multi-AI collaborative research platform coordinates 5+ AI models to jointly answer questions, cross-verify information, and provide credible consensus answers

198GitHub
Description

The first time you need to enter the background settings API first. The background address is https://kea-research. Your boxname.heiyu.space/admin, and the password is admin. Configure the openrouter API or a custom API. Usage method: 1. Select mode in the upper left corner 2. Enter research questions in the dialog box 3. Observe the four-step process of multi-AI collaboration: - **Initial*: Each AI answers independently - **Refine**: AI improves each other answers - **Evaluate**: AI mutual scoring and review - **Synthesize**: The optimal AI generates the final consensus answer ##Functional characteristics - ** Multi-AI consensus **: Call 5+ AI models (OpenAI, Anthropic, Google, Mistral, xAI, OpenRouter, Ollama) at the same time to independently answer the same question - ** Cross-validation **: AI models review each other, score, and identify consensus points and dispute points - ** Four-step process **: Initial answer → improvement → mutual evaluation → comprehensive output - ** Research level **: Create sub-dialogues from arbitrary messages and explore specific topics in depth - ** Fact extraction **: Automatically extract atomic facts and mark controversial content - ** Visual Intelligence **: Support image analysis (screenshots, charts, photos) - **75 languages **: Global multi-language support - **TTS voice **: Support text-to-speech reading - ** Export function **: Markdown, HTML, JSON, plain text multi-format export - ** User management **: Create and manage user accounts in the administrator panel

Screenshots
Screenshot 1
Screenshot 2
Screenshot 3
Screenshot 4
Screenshot 5
Screenshot 6
Screenshot 7
App Information
Version
0.2.0
Package Size
60.33 KB
Image Size
1.11 GB
Updated
March 4, 2026
Source Code
KeaBase
Platform Support
PC
Keywords
AI collaborationmultiple modelsresearch toolsfact checkingcross-verificationOpenAI