IdeAI vs Lilly

ideAI is the privacy-preserving memory and context layer for humans and AI. You can't see your own cognitive patterns. Your AI can't remember yesterday. ideAI closes both gaps — an encrypted, on-device memory layer that you browse as a semantic atlas, and your AI tools access via MCP. One data layer. Zero trust required. FOR YOU: THE ATLAS Import your digital history — journal entries, Twitter, Reddit, Claude, ChatGPT conversations — and watch it compress into a navigable topology. Day summaries. Week and month views. Semantic collision graphs that show where your thinking clusters, diverges, and evolves across time. FOR YOUR AI: DATASHARD Give any AI you already use — Claude, ChatGPT, Cursor — secure access to your Atlas via MCP. Every conversation starts with your full context, not a blank slate. Your AI finally has a memory. ON-DEVICE AI INSIGHT Chat with an on-device AI that has full context of everything you've written. It surfaces temporal patterns, semantic collisions, and recurring themes across months and years. Not advice — structure. ZERO-KNOWLEDGE PRIVACY AES-256-GCM encryption. On-device processing via Apple Foundation Models and MLX. Your data stays on your device. No servers see your content. No model trains on your thoughts. DATA SOVEREIGNTY You own all data in portable formats. You control all processing. You can export, delete, or migrate without loss. The value is in the instrument, not data captivity. Free to start. No account required.

Lilly records, transcribes, and summarizes your meetings — entirely on your iPhone. No servers. No cloud. No account required. Everything happens on-device using advanced AI models: • Voice activity detection separates speech from silence • Speaker diarization identifies who said what • Transcription captures every word with timestamps • AI summarization writes your meeting notes automatically Your audio and transcripts never leave your phone. Not during processing, not ever. Lilly is for rooms where confidentiality is not optional — legal meetings, medical appointments, investor calls, sensitive 1-on-1s. FEATURES On-device AI All transcription and summarization runs locally. No account required. No data uploaded. Speaker identification Lilly detects and labels multiple speakers. Name them after the meeting — names apply across the full transcript and summary. Auto-generated notes After each recording, Lilly produces structured meeting notes in the same language as your conversation. Full transcript Browse the complete timestamped transcript, attributed to each speaker. Offline first After the one-time model download, Lilly works with no internet connection at all. PRIVACY No analytics. No telemetry. No crash reporting with content. The only network request Lilly ever makes is the one-time AI model download on first launch (~3.3 GB). Verify this yourself: iOS Settings > Privacy & Security > App Privacy Report. REQUIREMENTS Requires iPhone 16 or later (A18 chip) running iOS 18. First launch downloads ~3.3 GB of AI models. After that, fully offline.