5 AIs. 4 Steps. 1 Answer You Can Trust.

Multi-AI collaboration platform that combines responses from multiple AI models,
cross-validates information, and delivers verified, consensus-backed answers.

---

The Challenge

When you ask a question to different AI models, you often get different answers — each presented with equal confidence. There is no built-in mechanism to determine which response is most accurate or to identify potentially incorrect information.

KEA Research addresses this by orchestrating multiple AI models to work together, evaluate each other's responses, and surface only the information that achieves consensus.

---

See It In Action

---

Installation & Update

Requires Docker: Docker Desktop (Mac/Windows) or Docker Engine (Linux)

Same command installs or updates — it detects existing installation automatically.

Linux / Mac

curl -fsSL research.kea.sh | sh

Windows (PowerShell)

irm research.kea.sh | iex

Manual Install

git clone https://github.com/keabase/kea-research.git
cd kea-research
cp .env.example .env
# Edit .env with your API keys
docker compose up -d

Manual Update

cd kea-research && git pull && docker compose up -d --build

---

How It Works

KEA processes every question through a structured 4-step pipeline:

Step	Process	Purpose
1. Initial	Each AI answers independently	Gather diverse perspectives without bias
2. Refine	AIs see anonymized peer responses	Synthesize best ideas, improve answers
3. Evaluate	AIs rank and review each other	Identify consensus facts and flag disputed claims
4. Synthesize	Best-ranked AI creates final answer	Deliver verified, trustworthy response

---

Key Features

KEA Research Layers

• Create focused sub-conversations from any message
• Explore specific topics in depth without cluttering main chat
• Each research layer maintains its own independent research thread
• Learn more about Research Layers methodology

Notes per Answer

• Add personal notes to any AI response

Visual Intelligence

• Add images directly to chat context
• Models can analyze screenshots, diagrams, and photos

Multi-AI Consensus

• Combines responses from 5 AI providers(Can be more than 5, up to you)
• Supports OpenAI, Anthropic, Google, Mistral, xAI, and Ollama
• Each model contributes independent analysis

Fact Verification

• Automatic extraction of atomic facts from each response
• Cross-validation identifies points of agreement
• Disputed or unsupported claims are flagged and excluded

Transparency

• Full visibility into each step of the process
• See how each AI responded and evolved its answer
• Understand why specific information was included or excluded

Flexible Export

• Export complete research with all steps
• Multiple formats: Markdown, HTML, JSON, Plain Text
• Include or exclude metadata and timestamps

Customization

75 languages supported (click to expand)

🇦🇿 Azərbaycan	🇮🇩 Bahasa Indonesia	🇲🇾 Bahasa Melayu	🇮🇩 Basa Jawa
🇧🇦 Bosanski	🇦🇩 Català	🏴󠁧󠁢󠁷󠁬󠁳󠁿 Cymraeg	🇨🇿 Čeština
🇩🇰 Dansk	🇩🇪 Deutsch	🇪🇪 Eesti	🇺🇸 English (US)
🇦🇺 English (AU)	🇬🇧 English (UK)	🇮🇪 English (IE)	🇪🇸 Español
🇪🇸 Euskara	🇵🇭 Filipino	🇫🇷 Français	🇮🇪 Gaeilge
🇪🇸 Galego	🏴󠁧󠁢󠁳󠁣󠁴󠁿 Gàidhlig	🇳🇬 Hausa	🇭🇷 Hrvatski
🇮🇸 Íslenska	🇮🇹 Italiano	🇰🇪 Kiswahili	🇱🇻 Latviešu
🇱🇺 Lëtzebuergesch	🇱🇹 Lietuvių	🇭🇺 Magyar	🇲🇹 Malti
🇳🇱 Nederlands	🇳🇴 Norsk	🇵🇱 Polski	🇵🇹 Português
🇧🇷 Português (Brasil)	🇷🇴 Română	🇦🇱 Shqip	🇸🇰 Slovenčina
🇸🇮 Slovenščina	🇫🇮 Suomi	🇸🇪 Svenska	🇻🇳 Tiếng Việt
🇹🇷 Türkçe	🇳🇬 Yoruba	🇬🇷 Ελληνικά	🇧🇾 Беларуская
🇧🇬 Български	🇰🇿 Қазақша	🇲🇰 Македонски	🇷🇸 Српски
🇺🇦 Українська	🇬🇪 ქართული	🇦🇲 Հայերdelays	🇮🇱 עברית
🇸🇦 العربية	🇮🇷 فارسی	🇵🇰 اردو	🇪🇹 አማርኛ
🇧🇩 বাংলা	🇮🇳 ગુજરાતી	🇮🇳 हिन्दी	🇮🇳 ਪੰਜਾਬੀ
🇮🇳 ಕನ್ನಡ	🇮🇳 मराठी	🇮🇳 தமிழ்	🇮🇳 తెలుగు
🇲🇲 မြန်မာစာ	🇹🇭 ไทย	🇨🇳 中文 (简体)	🇹🇼 中文 (繁體)
🇭🇰 中文 (香港)	🇯🇵 日本語	🇰🇷 한국어

• Light and dark themes
• Custom avatars and chat backgrounds
• Text-to-speech capability

Administration

• Web-based admin panel
• Manage AI provider API keys
• User account management
• System configuration

---

Supported AI Providers

OpenAI · Anthropic · Google · Mistral · xAI · OpenRouter · OpenAI-compatible (Ollama, LM Studio, etc.)

---

Use Cases

Research & Analysis

• Literature reviews requiring multiple perspectives
• Fact-checking and verification tasks
• Complex questions with nuanced answers

Professional Decision Support

• Business strategy analysis
• Technical architecture decisions
• Risk assessment and evaluation

Education

• Understanding where AI models agree and disagree
• Learning critical evaluation of AI-generated content
• Exploring topics from multiple angles

---

The Name

The Kea is a parrot native to New Zealand, recognized as one of the most intelligent bird species. Known for problem-solving abilities, tool use, and collaborative behavior — qualities that inspired this platform's approach to AI orchestration.

---

_{KEA Research — Verified answers through AI collaboration}