Perplexity Personal ComputerComplete Guide & Setup Tips

AI desktop agent for autonomous research, goal execution, and multi-step workflows. Setup guide, features, and productivity tips.

Workflow Planner

Describe a goal and see how Perplexity PC would break it into autonomous steps

Quick Setup Guide

Step 1: Install

Download Perplexity Personal Computer from perplexity.ai/pc. Requires macOS 14 (Sonoma) or later. Apple Silicon recommended for best performance — Intel Macs work but agent execution is ~40% slower.

Step 2: Grant Permissions

The app needs Accessibility, Screen Recording, and Files permissions. Go to System Settings → Privacy & Security and enable all three. These are required for the agent to see your screen and control apps.

Step 3: Configure Safety

Set your approval preferences: "Ask before" sending emails, deleting files, or making purchases. You can adjust these later. Start with all approvals enabled — you can relax them as you build trust.

Step 4: First Workflow

Start simple: "Research [topic] and summarize the top 5 findings." Watch how the agent works — it'll show its reasoning, the sites it visits, and let you approve or redirect at each step.

Key Features

Goal-Based Execution

Instead of step-by-step commands, describe your end goal. Perplexity PC evaluates multiple reasoning paths, selects the best approach, and executes autonomously. It adapts if something fails — retrying with alternative strategies.

Deep Research Integration

Built-in deep web research across 100+ sources simultaneously. Unlike browser-based search, it cross-references findings, evaluates source credibility, and synthesizes information into actionable summaries.

App Control

The agent can see your screen and control any app via mouse, keyboard, and native APIs. It works with browsers, spreadsheets, email clients, design tools, terminals, and more. Screen awareness means it can read UI elements, not just click coordinates.

Audit Trails

Every action is logged with timestamps, reasoning, and screenshots. You can review exactly what the agent did, why it chose that approach, and undo any action. Audit logs are stored locally and never leave your machine.

Scheduling

Set workflows to run on a schedule — daily market research, weekly competitor monitoring, automated report generation. The agent runs in the background and notifies you when complete.

Top 10 Workflows

1. Competitive Research

"Research my top 5 competitors, compare their pricing, features, and recent updates. Create a spreadsheet with findings."

2. Email Triage

"Read my last 50 emails, categorize them by urgency, draft replies for the 5 most important ones, and summarize the rest."

3. Meeting Prep

"Research [person/company] I'm meeting tomorrow. Find recent news, LinkedIn updates, and create a one-page briefing document."

4. Content Creation Pipeline

"Research trending topics in [niche], outline 5 blog post ideas with SEO keywords, and draft the introduction for each."

5. Code Review & Docs

"Review the latest PR in [repo], check for security issues, and update the README with any new features."

6. Financial Tracking

"Check my subscriptions across email receipts, compile a list with monthly costs, and flag any price increases."

7. Job Market Analysis

"Search for [role] positions in [city], compare salary ranges, required skills, and company ratings. Create a ranked shortlist."

8. Travel Planning

"Plan a 5-day trip to [destination]. Find flights under $[budget], top-rated hotels near [area], must-visit places, and create a day-by-day itinerary."

9. Social Media Monitoring

"Monitor mentions of [brand/product] across Twitter, Reddit, and HN. Summarize sentiment and flag any urgent issues."

10. Weekly Report

"Compile this week's progress from [project tool], summarize key metrics from [analytics], and draft a weekly update email."

Comparison with Other AI Agents

FeaturePerplexity PCClaude Computer UseOpenAI CodexApple Intelligence
Autonomous GoalsYesPartialCode-focusedNo
Deep ResearchNativeNoNoNo
Screen ControlYesYesLimitedPartial
App IntegrationAll appsAll appsTerminalApple apps
Audit TrailFullPartialYesNo
SchedulingYesNoNoPartial
Safety ControlsExtensiveBasicSandboxedNative OS
PlatformMac (Win soon)Mac/Win/LinuxMac/Win/LinuxApple only
Price$20/mo (Pro)API usage$200/moFree

Pro Tips

Frequently Asked Questions

What is Perplexity Personal Computer?
Perplexity Personal Computer is an AI agent platform that runs on your Mac desktop. Launched April 17, 2026, it shifts from traditional chatbot Q&A to goal-based autonomous execution. Describe what you want accomplished, and the agent researches, plans, and executes multi-step workflows — controlling apps, files, and browser on your behalf with full audit trails and safety controls.
Is my data safe?
Yes. All processing happens locally on your machine. Audit logs, screenshots, and workflow data stay on your computer. Web research queries go through Perplexity's servers (same as regular Perplexity search), but your local files and screen content are never uploaded. You can verify this in the audit trail which shows all network requests.
Can I stop a workflow mid-execution?
Yes! Press Cmd+Shift+X or click the kill switch button at any time to immediately stop the current workflow. The agent will safely halt, save its progress, and show you exactly where it stopped. You can resume from that point or start over.
Does it work offline?
Partially. Local file operations and app control work offline. Web research and AI reasoning require an internet connection. The agent will notify you if a workflow step needs internet access and queue it for when you're back online.
What are the system requirements?
macOS 14 (Sonoma) or later. Apple Silicon (M1 or newer) recommended for best performance. Minimum 8GB RAM, 16GB recommended. 2GB free disk space. Intel Macs are supported but agent execution is approximately 40% slower.