What is ChatBetter?
ChatBetter was an AI research platform developed by Unsupervised that investigated whether interacting with multiple large language models simultaneously could yield more trustworthy results than relying on any single model. The platform enabled users to explore multi-model orchestration techniques where questions were intelligently routed to the most appropriate model for each specific task.
The system incorporated quality guardrails and automated processes where AI agents would judge and combine outputs from different models, grounding responses in source data to enhance reliability. Over 30,000 participants contributed to this research preview, generating insights that are now influencing the development of future AI tools focused on trustworthy output generation and agentic data analysis.
Features
- Multi-Model Orchestration: Intelligent routing of questions to the most appropriate LLM for each specific task
- Automated Judgment: AI agents that automatically evaluate and combine outputs from different language models
- Source Grounding: Responses are anchored in source data to enhance accuracy and trustworthiness
- Quality Guardrails: Built-in mechanisms to ensure reliable and consistent AI outputs
Use Cases
- Researching multi-model AI interaction patterns
- Comparing outputs from different large language models
- Developing trustworthy AI systems with quality controls
- Exploring agentic approaches to AI output evaluation
