How ClayBot Works
The feedback loop that makes ClayBot smarter over time.
What is ClayBot?
ClayBot is the virtual representation of Clay Cash and his 30 years of experience in legal technology, eDiscovery, data architecture, and process design. Ask it questions and, where possible, it will give you answers that make sense based on what Clay knows.
Important: Clay does not guarantee these answers are correct. This system is purely informational. There is no guarantee of accuracy. All responses should be validated by qualified professionals before being relied upon for business, legal, or technical decisions.
The Feedback Loop
You Ask a Question
Visitor types a question into ClayBot
ClayBot Responds
AI generates an answer using Clay's knowledge base and Relativity documentation
Question Evaluated
System evaluates for valuable themes, business relevance, and technical substance. Meaningful questions are saved; small talk is discarded.
Clay Reviews
Clay personally reviews conversations. Questions ClayBot couldn't answer well? Clay provides the real answer and trains the system.
ClayBot Gets Smarter
New knowledge is added to ClayBot's training. Future visitors get better, more accurate answers.
How Questions Are Evaluated
Not every message is saved. The system classifies each question in real time:
Saved
- Business inquiries and engagement interest
- Technical eDiscovery questions
- Questions about Clay's expertise or approach
- Novel problems or creative ideas
- Questions with substance and depth
Discarded
- Greetings ("hi", "hello")
- Acknowledgments ("ok", "thanks")
- Yes/no responses
- Small talk without substance
- Farewells ("bye", "see ya")
All saved questions are stored anonymously. No personal information is attached. See our Privacy Policy for full details.
The Technology
ClayBot is powered by:
- Anthropic's Claude API — The AI that generates conversational responses
- Clay's Knowledge Base — 30 years of expertise captured through structured interviews and stored in a vector database
- Relativity Documentation — Indexed RelativityOne help and developer docs for technical accuracy
- RAG (Retrieval-Augmented Generation) — When you ask a question, the system retrieves the most relevant knowledge before generating a response
Important Disclaimer
ClayBot is an informational tool. It provides guidance on technical and business-related topics based on Clay Cash's experience, but:
- No guarantee of accuracy — Responses are AI-generated and may contain errors
- Not professional advice — Nothing here constitutes legal advice, certified technical guidance, or a professional consultation
- Always validate — Important decisions should be verified with qualified professionals
- Informational only — Clay Cash assumes no liability for actions taken based on ClayBot's responses
See our Terms of Service for the full legal disclaimer.