Appearance
Platform Capabilities
GhostAI provides all the productivity features you expect from a modern AI chat product — with enterprise-grade privacy and security built into every layer.
Chat Experience
Conversation Management
- Smart Auto-Titles — Conversations are automatically named by AI based on content
- Rename Conversations — Double-click or use the edit icon to rename any conversation
- Search Conversations — Full-text search across all your chat history
- Delete & Clear — Remove individual conversations or clear all history
Message Features
- Timestamps — Every message shows relative and absolute timestamps
- Regenerate Response — Re-run any response with one click
- Edit & Resubmit — Edit any user message and get a new response
- Copy Messages — One-click copy on every message, code block, and data field
- Thumbs Up/Down — Rate responses to help improve quality
- Download Code — Save any code block as a file with correct extension
- Read Aloud — Listen to any assistant message via multi-provider TTS (OpenAI, ElevenLabs, browser fallback)
Starter Prompts
- Quick-start prompts for common tasks (email, summarize, brainstorm, explain)
- Industry-specific demo prompts for security, healthcare, finance, and development
AI Capabilities
Web Search
GhostAI can search the web for real-time information when answering questions. Search queries are PII-sanitized before being sent to search providers.
Image Generation
Generate images using multiple models from several providers directly in chat. Image prompts are privacy-protected to prevent PII leakage.
- Persistent Model Selection — Choose your preferred image model and it's saved across sessions
- Multiple sizes — Square, landscape, and portrait options
- Works everywhere — Available in both regular chat and comparison mode
See Image Generation for full documentation.
SaaS Data Connectors
Connect third-party SaaS tools (Rippling, Salesforce, Slack, Jira) to enrich the AI's memory with organizational context. Neither ChatGPT nor Claude offers enterprise SaaS data connectors with PII filtering, source attribution, and access control scoping. GhostAI is the only platform that ingests SaaS data while automatically filtering PII fields before the AI sees it.
Rich Content Rendering
- LaTeX Math — Full math equation rendering with KaTeX
- Mermaid Diagrams — Flowcharts, sequence diagrams, Gantt charts, and more
- Markdown — Full GFM support including tables, task lists, and syntax highlighting
- Code Blocks — Syntax-highlighted code with copy, download, and language detection
Voice Assistant Mode
Full-duplex voice conversations with privacy-first architecture. PII is never spoken aloud.
- Voice PII Shield —
vocalize()replaces PII tokens with category labels so sensitive data is never spoken - Multi-provider TTS — OpenAI, ElevenLabs, browser fallback with streaming and automatic failover
- Local-first STT — faster-whisper on-server processing; cloud Whisper fallback
- Full-screen voice conversation — Animated orb, dual transcript, barge-in support
- Read Aloud — Listen to any assistant message with one click
- Voice Activity Detection — Hands-free conversation with ambient noise calibration
- Three privacy levels — Maximum (local only), Standard (labels + cloud), Relaxed
- Voice biometric consent — BIPA/GDPR-compliant consent management
- Multi-language — Automatic detection from speech
- Mobile PWA — Push-to-talk, lock-screen controls, Screen Wake Lock
Personalization
Custom Instructions
Tell GhostAI about yourself and your preferences. Instructions persist across all sessions, surviving restarts and working across devices. Access via the ⋮ More menu in the chat header.
- About You — Context about your role, industry, and expertise
- Response Style — How you want responses formatted (concise, detailed, technical, simple)
- Combinable Presets — Click multiple presets to combine styles
- Auto-save — Modal closes automatically after saving; warns if you have unsaved changes
Prompt Library
Pre-built and custom prompts for common tasks. Access via the ⋮ More menu.
- Professional email writing, document summarization, code review, compliance checks, incident reports, and more
- Add your own custom prompts
Prompt Enhancement (Magic Wand)
Transform rough prompts into detailed, well-structured queries with one click.
- Context-aware — Uses your industry vertical, custom instructions, and conversation thread
- Vertical-specific — Healthcare prompts reference HIPAA terminology, legal prompts cite frameworks, finance prompts incorporate regulatory language
- Undo — Revert to your original text within 8 seconds
- Keyboard shortcut —
Cmd/Ctrl + Shift + E
See Prompt Enhancement for full documentation.
Display & Accessibility
- Light/Dark Mode — Toggle themes or follow system preference
- Incognito Mode — Conversations are never saved to history or the database
- Keyboard Shortcuts — Quick keys for new chat, enhance prompt, send, and more
- Mobile Optimized — Responsive layout for phones and tablets
Transparency
Token & Cost Visibility
Every response shows exact token counts, per-message credit cost, context window usage, response time, and model used. Account-level usage is available in the sidebar with credit breakdowns, trend sparklines, and CSV export.
Context Composition (Team+)
The context usage bar shows a segmented breakdown of how the context window is allocated: system prompt, memory, attachments, history, and tools. Per-message savings indicators show how many tokens were optimized away and the overall efficiency. See Context Optimization for details.
API Key Usage & Cost Tracking
Track per-provider API usage, token consumption, estimated costs, and error rates. See Usage & Cost Tracking for details.
Extended Thinking
When models think before responding, see the thinking duration and expand to view reasoning.
What's New
In-app changelog so you never miss new features. Badge indicator when updates are available.
Model Comparison
Side-by-Side Comparison
Compare 2–4 AI models with the same prompt. Responses stream in real-time with full PII protection.
- File attachments — Upload documents and images with automatic PII scanning
- Web search & image generation — Toggle tools per-model or shared across models
- Incognito mode — Comparisons that are never saved
- Real-time PII preview — See detected PII while typing with in-context highlighting
- Per-model retry — Retry a single failed model without restarting
- Vote & promote — Pick a winner and continue as a regular chat
- Export as Markdown — Download full comparison results
Consensus Synthesis
After all models respond, generate a structured analysis covering agreements, contradictions, and a recommended answer. Iterate on the consensus with follow-up questions.
Combined Answer — Multi-Model Deliberation
Run multiple models in parallel, then blend their responses into a single, refined answer. Also available in single chat via the "Combined Mode" toggle.
Mega — Cross-Round Intelligence
Work across multiple comparison rounds to synthesize or combine insights from earlier conversations into a unified analysis.
Privacy-First Differentiators
GhostAI goes beyond standard AI chat with privacy and security capabilities that are absent from consumer products:
- Automatic PII Detection — Broad entity coverage across names, addresses, financial data, health information, and more
- Compliance Tagging — HIPAA, PCI, GDPR, SOX, and other regulatory frameworks
- Token Vault Inspection — See exactly what the AI model receives after privacy processing
- Source Code Protection — Detect and block accidental code leakage
- Image PII Redaction — Automatically redact faces, IDs, and cards from uploaded images
- File Intelligence — Semantic search with PII-aware retrieval across uploaded documents, supporting a wide range of file types
- File Threat Inspection — Multi-layered scanning for malware, embedded scripts, prompt injection, and other document-borne threats
- Post-Response PII Guard — Scans model output for PII before delivery
- Tool Argument Validation — Blocks dangerous patterns in model-generated tool calls
- Security Download Guard — Risk-tiered download controls with full audit trail
- Stream Resilience — Auto-retry with backoff and partial response preservation
- AI Memory — Multi-scope memory hierarchy with organizational knowledge bases and PII-protected storage
- BYOK (Bring Your Own Key) — Use your own API keys for full control over model access and billing
- Audit Logging — Comprehensive activity logs for enterprise compliance
- Investor Dashboard — Observer role with shareable links for stakeholders