← Back
VentureBeat AI

Your files stay put: Perplexity’s hybrid AI keeps confidential data off the cloud

•8 min read•
#deployment#agents#llm#compute
Your files stay put: Perplexity’s hybrid AI keeps confidential data off the cloud
Level:Intermediate
For:ML Engineers
✦TL;DR

Perplexity’s new hybrid compute feature for its agentic platform, Computer, lets a single AI agent split its workload between powerful frontier models in the cloud and smaller open‑weight models running locally on Apple silicon Macs. By routing sensitive data to the local machine, the system keeps confidential information off the cloud while still leveraging cloud‑scale compute for non‑sensitive tasks. The approach demonstrates a practical trade‑off between data privacy and the raw performance of large models, but it requires Apple silicon hardware and a local deployment of the open‑weight model.

⚡ Key Takeaways

  • Hybrid compute splits tasks between cloud frontier models and local open‑weight models on Apple silicon Macs.
  • The agentic platform “Computer” routes sensitive data exclusively to the local model to preserve privacy.
  • Engineers must provision Apple silicon Macs and deploy the local open‑weight model to use the feature.
  • The solution is limited to Apple silicon hardware; other platforms are not supported yet.
  • WhyItMatters: This design lets production systems keep private data on-premises while still harnessing cloud‑scale AI, reducing compliance risks for sensitive workloads.
  • TechnicalLevel: Intermediate
  • TargetAudience: ML Engineers
  • PracticalSteps:
  • Deploy the open‑weight model on an Apple silicon Mac and configure it as a local endpoint in the Computer platform.
  • In the Computer workflow, specify routing rules that direct sensitive data to the local endpoint.
  • Monitor latency and resource usage to balance privacy with performance.
  • ToolsMentioned: Apple silicon Macs, Perplexity Computer
  • Tags: DEPLOYMENT, AGENTS, LLM, COMPUTE

🔧 Tools & Libraries

Apple silicon MacsPerplexity Computer
💡 Why It Matters

This design lets production systems keep private data on-premises while still harnessing cloud‑scale AI, reducing compliance risks for sensitive workloads.

✅ Practical Steps

  1. Deploy the open‑weight model on an Apple silicon Mac and configure it as a local endpoint in the Computer platform.
  2. In the Computer workflow, specify routing rules that direct sensitive data to the local endpoint.
  3. Monitor latency and resource usage to balance privacy with performance.

Want the full story? Read the original article.

Read on VentureBeat AI ↗

More like this

Censored vs. Truncated Regression in Python, Explained

Towards Data Science•#llm

New in AWS Quick: More capable flows, an updated web experience, and agents that work together

AWS ML Blog•#agents

Three insights you may have missed from theCUBE’s coverage of the Dell Technologies New York Symposium

SiliconANGLE AI•#deployment

How Google Duplex evolved to also call businesses for you in Gemini

Google Research•#agents

EXPLORE AI NEWS

Daily hand-picked stories on LLMs, RAG, agents and production AI — curated for engineers who ship.

BROWSE NEWS

GET THE WEEKLY DIGEST

Join engineers getting the Monday signal-over-noise AI breakdown. No spam, unsubscribe anytime.

LEARN AI ENGINEERING

Curated courses, research papers, repos and tutorials built for engineers leveling up in AI.

START LEARNING