Show HN: Riverse – persistent AI memory that grows with you, no RAG (github.com)

Riverse, a groundbreaking personal AI agent, has been unveiled, designed to operate on users' own devices with a focus on persistent memory and offline cognition. Unlike existing AI models that reset knowledge with each platform change, Riverse consolidates memories from conversations in real-time, creating a personalized profile that evolves over time. The agent employs a unique River Algorithm to solidify insights into a timeline-based structure, resembling the natural flow and sedimentation processes of a river, thereby helping the AI to reinforce confirmed knowledge and discard contradictions. This innovation is significant for the AI/ML community as it addresses the limitations of current memory systems used in AI, which often rely on cloud storage and provide shallow, disconnected data points. Riverse incorporates multi-modal inputs, allowing interaction through various channels such as Telegram and Discord, and offers features like YAML-based custom skills and proactive outreach. As a local-first solution built on technologies like PostgreSQL and Python, it emphasizes data ownership and privacy, making it particularly relevant amid growing concerns over data security in AI applications. The beta version is tailored for individual users, suggesting a promising shift towards personalized and secure AI interactions.