Hello Haven Debuts Personal AI Twin with $15 Million Funding

Unlike standard large language models that require constant context-setting, Haven learns from every interaction to build a unique history of a user’s preferences, relationships, and decision-making patterns. By operating as a personal layer that sits above various AI models and apps, the system can autonomously triage emails, place orders, or handle scheduling across phone, text, voice, and web interfaces. Since entering a private beta phase last month, the platform has already logged over 2 million user interactions.

Founder Frank Addante describes the technology as a shift from reactive assistance to proactive execution. Because the system maintains a user's context in a private memory graph—rather than locking it into a specific model—the digital twin is designed to remain consistent across different devices and future hardware, including televisions and automobiles. Navin Chaddha of Mayfield, who previously backed Addante at Rubicon Project, noted that the product represents a pivot toward AI that retains individual judgment rather than merely answering general queries. Haven is available immediately through web, iOS, and Android platforms, with plans to open a developer ecosystem for third-party integrations.