Prifina

Prifina is a persistent personal intelligence layer beneath AI products. It keeps context, memory, permissions and sharing boundaries around the individual while different models do the work, and it powers Souvéa and AI Twin, both live today.

Prifina is building what it calls a persistent personal intelligence layer: a foundation for personal context, memory, permissions, relationships and sharing boundaries that sits around the person rather than inside any single AI service. The premise is straightforward. Models get swapped, agents multiply and applications come and go, yet personal context is repeatedly rebuilt inside each new product. Prifina's answer is to keep that context in one persistent space below the model, so the intelligence can change without the person starting over. The layer has four parts, all described as available today. A personal data space holds files, messages, calendar, health, contacts, travel, plans and personal knowledge. AI choice lets different models and capabilities (planning, research, health and wellbeing, communication) work against that space without rebuilding it. Storage and processing choices let a person decide whether data and AI requests run on a personal device, a trusted cloud, private infrastructure or systems in a chosen region. Shared contexts let selected information be contributed to a group or organisation environment while the rest stays private. In practice most people meet the architecture through two products. Souvéa is the consumer side: a personal assistant on iPhone and web that connects chosen information, summarises overnight messages, prepares a day plan from calendar and health data and asks for approval before sharing details. Prifina's AI Twin is the professional side: an approved, cited version of a person's knowledge that answers within scope and declines questions outside it. Spaces and Zones extend the same model to groups and organisations. The tool suits privacy-conscious professionals, academics, consultants and organisations that want AI assistance without handing a single vendor the full picture. Its natural comparison point is a platform assistant that keeps context locked inside one ecosystem. Prifina's difference is architectural rather than a feature list, and the trade-off is that setup asks more deliberate choices about storage, permissions and sharing than most consumer AI apps do.