Private Data AI Chat

Clear Ideas integrates leading AI models including OpenAI (GPT-4, GPT-4 Turbo, GPT-4o), Anthropic (Claude 3.5 Sonnet, Claude 3 Opus), Google (Gemini 1.5 Pro), Cohere, and xAI (Grok). The Intelligent model selector automatically chooses the optimal model based on your task complexity, content type, and cost requirements. You can also manually select specific models for different use cases.

AI credits are consumed based on the amount of data processed (input tokens from your prompts and retrieved content) plus generated output (response tokens). Different models have different credit rates—advanced reasoning models consume more credits than standard chat models. Your subscription includes a monthly AI credit allocation, and additional credits can be purchased.

No. Clear Ideas has zero-retention agreements with all AI model providers. All data transmission is encrypted, and only the minimum necessary context is shared with models to generate responses—your full dataset never leaves the secure Clear Ideas environment.

AI Chat respects all permission controls and site-scoped security. When you start a chat conversation, you select which sites the AI can access. The AI retrieves only from documents within those selected sites that you have permission to view. This ensures external users cannot query content outside their authorized access, and conversations remain within appropriate security boundaries.

AI Chat responses are grounded in your actual private data with source citations for every claim. The retrieval system finds relevant content from your documents and provides it as context to the AI model, which generates responses based solely on that verified content. Citations link back to specific documents so you can verify accuracy. Deterministic grounding in private data eliminates hallucinations common in generic AI chatbots.

Yes, use Output Templates to guide AI responses toward specific formats, tones, or content structures. Templates are markdown documents stored in your sites that define how the AI should structure responses. For example, create templates for executive summaries, technical documentation, compliance reports, or client-facing communications. Each chat can reference different templates for consistent output formatting.

Yes, save any chat conversation to your sites as a permanent record, copy responses to clipboard for pasting elsewhere, or export entire conversations as beautifully formatted PDFs with all source citations intact. PDF exports are perfect for sharing AI-assisted analysis with stakeholders, creating documentation of research processes, or archiving important conversations for compliance purposes.

Model Context Protocol (MCP) allows compatible AI applications (like Claude Desktop, Cursor IDE) to securely query your Clear Ideas content during their workflows. When enabled, these applications can retrieve relevant context from your private data to enhance their responses—all while maintaining Clear Ideas security controls. Only authorized applications with valid access keys can connect, and all interactions are logged in audit trails.

Yes, fork any message in a conversation to create a new branch exploring alternative approaches or questions. This is perfect for comparing different AI models on the same question, testing various prompt phrasings, or exploring multiple analytical directions without losing your original conversation thread. Each fork maintains its own independent history while preserving the shared conversation context up to the fork point.