This is our most ambitious project yet.
At vaic.at, we’ve always believed that intelligence should complement you, not monitor you. Today, we’re thrilled to unveil the first details about a new layer in our vAI architecture: Private Compute Cubes (PCC) which is now undergoing active development and internal alpha testing.
A New Paradigm: Private Compute Cubes
Private Compute Cubes, or PCCs, are designed to function as temporary, privacy-preserving AI proxies. These proxies collect structured, context-relevant information online, such as housing data, job listings, or salary info, without exposing any personal user data.
Each PCC is a short-lived, local simulation of a generic user. This user is trained to safely gather information from public or licensed data sources. Once its task is complete, the PCC securely deletes itself, leaving no trace, profile, or persistent identifiers behind.
In essence, PCCs operate in a one-way manner: they observe the internet, but nothing returns to them.
How PCC Works
At its core, PCC operates locally using large language models (LLMs) to simulate browsing patterns through randomized, anonymized profiles that cannot be linked to any real person.
Our systems exclusively retrieve data from legitimate, permissioned sources. The results are then transmitted to CathAI and CatharinAI for analysis.
Users have the option to review the raw data, sources, and insights retrieved, ensuring complete transparency and trust.
During alpha testing, PCC utilizes locally deployed MistralAI models, enhanced with vaic.at’s proprietary NLP stack. This setup allows for controlled experimentation with contextual inference, all within a privacy-preserving, on-device environment.
Building for the Future of Apple Intelligence
As Apple continues to expand App Intents, Apple Intelligence, and Foundation Models, we anticipate significant potential for PCC to seamlessly integrate into the Apple ecosystem.
Our long-term objective is to make our AI features accessible to Siri and Apple Intelligence once official APIs are released, ensuring that privacy-first computation remains the cornerstone of every interaction.
Why This Matters
Across our ecosystem, from mEUvy to EUnify, EUorigin, and Salary Insights, AI plays or will play a supportive role. It provides context, surfaces insights, and helps users make informed decisions.
With PCC, we’re extending this principle beyond the app. AI can now gather relevant information for you without ever compromising your privacy.
PCC represents a new model for how AI can exist within strict privacy boundaries.
What’s Next
Alpha testing of PCC has commenced within CUEB.ES, and early prototypes are currently being evaluated for integration with CathAI and CatharinAI.
Further details regarding developer collaboration, open testing, and regulatory compliance documentation will be shared in late 2026.
Closing Thoughts
At vaic.at, we’re not building AI that replaces people we’re building intelligence that protects them.
Private Compute Cubes are one more step toward that vision: empowering users to benefit from AI-driven insights without sacrificing privacy, autonomy, or trust.
PCC is in alpha testing. Specifics about this project can still change.
Comments
There are no comments yet.