Building Your Local AI Workspace: A Complete Guide to Perplexity's Mac-Native Personal Computer Platform

Overview

In a recent earnings call, Apple spotlighted Perplexity’s innovative Personal Computer platform, calling it a “game-changer” for privacy-first AI computing. Perplexity has responded by revealing deeper technical details about their Mac-native solution. This guide walks you through everything you need to know: what the platform is, how to set it up on your Mac, and how to avoid common pitfalls. Whether you’re a developer, researcher, or curious power user, you’ll learn to harness a fully local AI assistant that keeps your data on-device—no cloud required.

Prerequisites

Before diving in, ensure your system meets these requirements:

• Mac hardware: Apple Silicon (M1, M2, M3, or M4 series) or Intel-based Mac with at least 16 GB unified memory (32 GB recommended for larger models).
• macOS version: Ventura 13.4 or later (Sonoma 14+ preferred for Metal 3 acceleration).
• Storage: At least 10 GB free space for the base installation; extra 30–50 GB if you plan to download multiple language models.
• Internet connection: Required only for initial download and updates—the platform runs completely offline thereafter.
• Apple ID: Needed to verify your Mac and receive platform updates via the Mac App Store (or direct installer).

Step-by-Step Instructions

1. Download and Install the Platform

1. Open the Mac App Store and search for “Perplexity Personal Computer”. Alternatively, visit the official Perplexity website and download the .dmg installer.
2. Double-click the installer and drag the Perplexity PC app into your Applications folder.
3. Launch the app. You’ll be prompted to grant permissions for Microphone (voice input), Accessibility (system-wide search hotkey), and Files & Folders (document indexing). Accept all to unlock full functionality.
4. On first launch, the platform downloads a default lightweight model (~2 GB) optimized for your Mac’s chip. This takes 5–10 minutes on a standard broadband connection.

2. Configure Your Local Compute Engine

After installation, open Perplexity PC > Preferences and adjust these settings:

• Model Selection: Choose from available open-source models (e.g., Mistral 7B, Llama 3, Perplexity’s custom pplx-7b). Each option shows estimated RAM usage and speed on your Mac.
• Compute Mode: Select High Performance (uses all CPU/GPU cores) or Eco Mode (limits to 50% for lower power draw).
• Storage Path: Designate a folder for models and cached data. Default is ~/Library/Application Support/com.perplexity.pc/.

Pro tip: If you have an external SSD, point the storage path there to save internal space.

3. Index Your Documents for On-Device AI

The platform can read and understand your local files—PDFs, Word docs, Notes, emails (via Apple Mail), and code repositories. To enable this:

1. Go to Preferences > Document Indexing.
2. Click the + button and select folders you want to index (e.g., ~/Documents, ~/Desktop).
3. Choose indexing depth: Quick (metadata only) or Deep (full text + embeddings). Deep indexing takes longer but enables rich Q&A.
4. Click Start Indexing. The process runs in the background and uses negligible CPU once embeddings are generated.

4. Running Your First Query — Voice or Text

You can interact in two ways:

• Text mode: Use the floating window (invoked via Cmd+Space, then type your question). Example: “Summarize the Q2 earnings call transcript I saved yesterday.”
• Voice mode: Press the microphone icon in the menu bar, then speak. The platform uses on-device speech recognition (powered by Apple’s Speech framework). No audio leaves your Mac.

A sample workflow: Ask “What are the key points from my notes on event-driven architecture?” The AI will retrieve relevant snippets from indexed notes and generate a bullet-point answer—all offline.



5. Advanced: Custom Actions and Automation

For power users, Perplexity PC supports Shortcuts and AppleScript to chain actions. For example:

• Create a shortcut that asks the AI to “read selected text and reply with a simplified version” and assign it a keyboard shortcut.
• Use the perplexity command in Terminal with arguments: perplexity --query "What is 2+2?" --model pplx-7b.

See the official Common Mistakes section below for troubleshooting integration issues.

Common Mistakes

Mistake 1: Not Enabling Full Disk Access

If the platform can’t find your files, you likely declined Full Disk Access during setup. Go to System Settings > Privacy & Security > Full Disk Access and toggle on Perplexity PC. Without this, document indexing fails silently.

Mistake 2: Overloading the Model with Insufficient RAM

Selecting a 70B-parameter model on a 16 GB Mac will cause the app to crash or swap aggressively. Always check the RAM indicator in Preferences. A good rule: 7B models need ~8 GB; 13B models need ~16 GB; 34B+ models require 32 GB or more.

Mistake 3: Forgetting to Update the Core Engine

Perplexity frequently releases performance patches and new model support. If you see a red dot on the menu bar icon, open Perplexity PC > Check for Updates. Staying current ensures compatibility with macOS updates.

Mistake 4: Indexing Too Many Folders at Once

Selecting the entire /Users/ directory can cause the indexing to take hours and consume battery. Limit to specific project folders. Use the Exclude list to omit system folders like Library/Caches.

Summary

Perplexity’s Personal Computer platform brings the power of large language models directly to your Mac, with zero data leaving your device. By following the setup steps—installation, configuration, document indexing, and querying—you can create a private, always-available AI assistant. Avoid the common pitfalls of insufficient RAM, missing permissions, and over-indexing to keep the experience smooth. With Apple’s endorsement, this platform marks a significant step toward truly personal, local AI computing.