Thesis6 min read
Why AI's next era needs a personal intelligence layer
Every service that rebuilds a person's context creates another dependency. A persistent layer removes the need to rebuild it at all.
Read the article →INSIGHTS
Defining the architecture of persistent personal intelligence.
Working arguments, product evidence and technical questions around context, memory, permissions, model independence, shared intelligence and governed AI.
What is different: these are arguments and open questions, not product announcements. What to do next: read the thesis for the full case.
PUBLISHED
Latest writing.
Research5 min read
From centralized data to intelligence that learns with individuals
The first era learned from pooled records. The next level may require learning alongside a person whose record never moves.
Read the article →Technology5 min read
Where should a personal AI request be processed?
Storage and processing are different choices. Treating them as one hides the decision that matters most.
Read the article →Company4 min read
How Souvéa and AI Twin test the Prifina foundation
Prifina builds the foundation. Souvéa works for you. AI Twin represents you. Two products, one set of principles.
Read the article →
Company4 min read
Prifina and Mastercard Lead the Way in Personal AI Innovation
Prifina brought Personal AI Twins to Mastercard's LAC Innovation Forum 2025 in Miami, giving Mastercard leadership and invited executives direct access to the technology.
By Peter Steube
Read the article →IN PREPARATION
What we are writing next.
These pieces are drafted and awaiting publication. We do not attach an author or a date to anything until it is actually published.
ThesisIn preparation
Model choice means little without control of the data relationship
Swapping models is easy. Swapping the platform that holds your history, permissions and relationships is not.
TechnologyIn preparation
Shared context: collaboration without dissolving the personal boundary
A group can hold shared context without any member handing over their whole personal context.
TechnologyIn preparation
Why governed AI agents need permissions and provenance
An agent that cannot show its scope or its sources cannot be trusted with a person's context.
NEXT