Show HN: Memno – AI with persistent memory that runs in an isolated environment

Hacker News - AI
Jul 16, 2025 12:41
blaget
1 views
hackernewsaidiscussion

Summary

Memno is an AI assistant designed as a "chief of staff" that securely stores and indexes user data—including photos, documents, and voice notes—in an isolated, encrypted environment, enabling fast, context-aware queries and actions like making calls or sending texts. Each user receives a dedicated vector database and storage, ensuring complete data isolation and privacy. This approach highlights advances in personalized, privacy-focused AI assistants with robust memory and rapid search capabilities, addressing key concerns around data security and user control in AI applications.

We built Memno. We like to think of Memno as an AI chief of staff. Every photo, document, and voice note you share becomes searchable memory. Ask "What did we decide about Q4?" and it pulls from all your sources. Can also make calls and send texts on your behalf (with approval!) Each user gets their own isolated environment with dedicated vector database, metadata storage, and object storage instances. Some highlights: - Complete data isolation per user (no shared databases), encrypted at rest - Sub-50ms vector search - Document processing pipeline for OCR/transcription/object recognition/summarization/etc. before vectorization. - Distributed AI calls (no provider sees everything) Comments URL: https://news.ycombinator.com/item?id=44581642 Points: 1 # Comments: 0