Cookie & Data Capabilities Policy
Last updated: February 24, 2026
The Short Version
This site uses zero tracking cookies. No advertising cookies. No analytics cookies. No third-party tracking scripts. We use minimal browser storage for functionality only, and we're transparent about exactly what data is captured from chat conversations.
Browser Storage Used
Instead of cookies, this site uses browser storage APIs for two specific, functional purposes:
| Storage Type | Key | Purpose | Duration |
|---|---|---|---|
| sessionStorage | claybot_session | Unique session identifier (UUID) for grouping chat messages within a single browser session. Used to track conversation continuity, not to identify you personally. | Until tab/window closes |
| localStorage | claybot_consent | Records whether you've dismissed the consent banner. Stores only the value "accepted" — no personal data. | Persistent (until cleared) |
Neither of these stores personal information, and both can be cleared at any time through your browser settings.
Chat Question Capture — How It Works
When you chat with ClayBot, every question you send is evaluated for substance in real time. Here's the transparent breakdown:
Questions That ARE Stored
Substantive, interesting, or business-relevant questions are stored anonymously. This includes:
- Business inquiries (e.g., "Can you help our law firm with eDiscovery?")
- Hiring or engagement interest (e.g., "What are your rates?")
- Partnership or collaboration proposals
- Specific project requests
- Thoughtful technical or professional questions
- Questions with unique perspectives, ideas, or novel problems
- Messages that contain contact information you voluntarily provide
Questions That Are NOT Stored
Simple conversational messages are evaluated and discarded immediately. This includes:
- Greetings ("hi", "hello", "hey")
- Acknowledgments ("ok", "thanks", "cool", "got it")
- Yes/no responses
- Farewells ("bye", "see ya")
- Generic small talk with nothing unique or idea-based
What Gets Stored With Each Question
| Data Point | Description | Why |
|---|---|---|
| Question text | Your message as written | To understand what visitors are asking about |
| Bot response | First 500 characters of ClayBot's reply | Context for understanding the conversation |
| Category | Auto-classified topic (e.g., "business_inquiry", "technical_ediscovery") | Trend analysis and organizing questions |
| Timestamp | When the question was asked | Chronological ordering |
| Session ID | Random UUID from your browser session | Grouping questions from the same conversation |
| Hashed IP | SHA-256 hash of your IP, truncated to 16 characters | Rate limiting and abuse prevention. Your actual IP is never stored. |
| Substance score | Numeric score from our classification system | Evaluating question quality |
| Conversation length | Number of user messages in the current conversation | Understanding engagement depth |
| User agent | Your browser's user-agent string, truncated to 200 characters | Understanding what devices and browsers visitors use |
| Flagged status | Whether the question was flagged as high-interest or high-engagement | Prioritizing review of actionable inquiries |
Server-Side Session Tracking
When you chat with ClayBot, a temporary session record is created on our server containing your message count, first and last seen timestamps, and hashed IP address. This data automatically expires after 24 hours. It is used to understand conversation engagement depth — not to identify you personally.
High-engagement flagging: If you send 5 or more messages in a session, subsequent substantive questions are automatically flagged as "high engagement" to help Clay prioritize review. No additional personal data is collected.
Documentation Search (RAG)
ClayBot uses a vector database (Upstash Vector) containing pre-indexed RelativityOne documentation to provide more accurate answers about eDiscovery topics. When you ask a question, it is sent as a search query to find relevant documentation excerpts. Your questions are not stored in the vector database — it contains only publicly available documentation.
Third-Party Services
The following services are involved in the site's operation. Each has their own privacy and cookie policies:
| Service | Purpose | Data Involved |
|---|---|---|
| Vercel | Hosting & serverless functions | Standard web request data (IP address, user agent) for serving pages. Vercel may set functional cookies for their infrastructure. |
| Anthropic (Claude API) | AI chatbot responses | Your chat messages are sent to Anthropic's API to generate responses. Subject to Anthropic's data handling policies. Messages are processed in real time and not retained by Anthropic for training per their API terms. |
| Upstash Redis | Question storage, session tracking, rate limiting | Anonymized question data, session records (24h TTL), and rate-limiting counters. Hosted in the US. Encrypted at rest and in transit. |
| Upstash Vector | Documentation search (RAG) | Pre-indexed RelativityOne documentation. Your questions are used as search queries but not stored. No user data in this database. |
| Google Fonts | Font delivery | Your browser makes requests to Google's CDN to load fonts. Google may log standard request data (IP, user agent). No cookies are set by this service on our site. |
Data Retention
- Stored questions: Retained indefinitely for trend analysis unless you request deletion.
- Session data (server-side): Automatically deleted after 24 hours.
- Daily statistics: Retained for 90 days, then automatically deleted.
- Browser storage: Session storage clears when you close the tab. Local storage persists until you clear it.
Your Controls
- Clear browser storage: Delete sessionStorage and localStorage for this site via your browser's developer tools or settings.
- Request data deletion: Email info@cdiscovery.io to request deletion of any stored questions. Since questions are stored with hashed session IDs (not personal identifiers), provide the approximate date/time and content of your questions to help us locate them.
- Opt out of question storage: There is no opt-out mechanism for question evaluation, as it is integral to the service. If you prefer not to have your questions evaluated, please contact Clay directly at info@cdiscovery.io instead of using the chatbot.
Contact
Questions about our cookie and data practices? Contact us at info@cdiscovery.io.