# I Gave AI My Medical Records and a Parking Ticket **URL:** \ **Category:** Content Creators\ **Tags:** alex-ziskind, security, tech, running-locally\ **Created:** [20 October 2026 17:30 UTC](https://www.artofsm.art/t/i-gave-ai-my-medical-records-and-a-parking-ticket/26505 "2026-10-20T17:30:03Z")\ **Posts on this page:** 2\ **Page:** 1
**Author:** ![artesia](https://www.artofsm.art/user_avatar/www.artofsm.art/artesia/32/36_2.png) [@artesia](https://www.artofsm.art/u/artesia)\ **Post date:** [20 October 2026 17:30 UTC](https://www.artofsm.art/t/i-gave-ai-my-medical-records-and-a-parking-ticket/26505/1 "2026-10-20T17:30:03Z")
[![](https://www.artofsm.art/uploads/default/original/3X/8/4/84b3f3e7474f539f8deeb644f538eae22f9abda0.jpeg "I Gave AI My Medical Records and a Parking Ticket") ](https://www.youtube.com/watch?v=8FEm8mMzeLE) The video showcases Perplexity’s hybrid AI compute app, which balances privacy and performance by processing sensitive data like medical records and financial documents locally on a Mac while leveraging cloud AI for less sensitive tasks. This approach ensures user control over private information through features like a privacy gate and local text recognition, offering a secure and efficient solution for personal AI use without compromising data confidentiality. ---
**Author:** ![artesia](https://www.artofsm.art/user_avatar/www.artofsm.art/artesia/32/36_2.png) [@artesia](https://www.artofsm.art/u/artesia)\ **Post date:** [20 October 2026 17:50 UTC](https://www.artofsm.art/t/i-gave-ai-my-medical-records-and-a-parking-ticket/26505/3 "2026-10-20T17:50:21Z")
The video explores the use of hybrid AI computing to balance privacy and performance when handling sensitive data like medical records and financial documents. The creator expresses a desire to use AI for personal tasks without sending private information to the cloud, highlighting the limitations of local AI models in terms of capability. The solution presented is a hybrid approach that combines local AI processing on a Mac with cloud-based AI, allowing sensitive data to stay on the device while leveraging the cloud for more complex tasks. This approach is facilitated by Perplexity’s new hybrid compute app, which automatically decides which parts of a task to process locally and which to send to the cloud. The presenter tests the Perplexity hybrid compute app on three different Macs with varying hardware capabilities, demonstrating how the app selects appropriate local AI models based on the machine’s RAM and processing power. The app includes a “privacy gate” feature that detects and filters out private information before sending any data to the cloud, ensuring that sensitive details like bank statements, medical records, and parking tickets remain on the local device. The presenter shows how the app flags private data and asks for user confirmation before processing, emphasizing the importance of user control over data privacy. Using sample data, including internet service agreements, bank statements, and bills, the presenter tests the app’s ability to match payments to bills and flag discrepancies while keeping financial documents private. The hybrid system successfully processes non-sensitive information in the cloud, such as web research for better internet deals, while keeping personal financial data local. The presenter also highlights the synchronization feature that keeps conversations consistent across devices, adding convenience to the hybrid workflow. The video further demonstrates the app’s capability to handle images containing sensitive information through local text recognition, preventing private data from being uploaded to the cloud. Examples include analyzing a car rental down payment, medical records, and a parking ticket. The app processes these images locally and only sends non-private, redacted information to the cloud when necessary. This selective sharing ensures that highly sensitive data never leaves the user’s Mac, maintaining privacy without sacrificing AI functionality. In conclusion, the hybrid AI approach showcased in the video offers a practical balance between convenience and security. By intelligently splitting tasks between local and cloud models, users can benefit from powerful AI assistance while keeping their most sensitive information private. The presenter praises the seamless integration and automatic privacy protections of Perplexity’s hybrid compute app, encouraging viewers interested in privacy-conscious AI to try it out. The video underscores the importance of hybrid AI as the future of secure and efficient personal data processing. ## Useful Links - [Perplexity Hybrid Compute](https://www.perplexity.ai/hub/products/hybrid-compute) — Directly enables the hybrid AI approach discussed in the video. - [PII Tracer by Perplexity AI on Hugging Face](https://huggingface.co/perplexity-ai/PII-Tracer) — Substantiates the privacy gate feature that filters private data before sending to the cloud. - [Gemma 4 E4B Open Weights Model](https://ai.google.dev/gemma/docs/core) — Explains one of the local AI models used in the hybrid compute app. - [Quen 3.635B Open Weights Model](https://huggingface.co/Qwen/Qwen3.6-35B-A3B) — Details another local AI model option for the hybrid approach. - [Perplexity AI Official GitHub Gist for Silicon Exchange Site Prompt](https://gist.github.com/alexziskind1/fe55f03892f3fe1d6d8cf6065c631bb8) — Provides the prompt configuration used in the hybrid compute app demonstration.