What Happened
On March 11, 2026, inside a converted church in San Francisco's North Beach neighborhood, Perplexity held its first developer conference — Ask 2026 — and announced Personal Computer. The product is software, not hardware: a persistent AI agent that runs continuously on a user-provided Mac mini, giving it direct access to local files, applications, and ongoing sessions.
Perplexity CEO Aravind Srinivas framed the vision plainly: “A traditional operating system takes instructions; an AI operating system takes objectives.” Personal Computer is the attempt to build that second kind of system on top of macOS.
Key Points
- Always-on local agent: Personal Computer runs 24/7 on a dedicated Mac mini, accessing your local file system and macOS applications — not just web-based services
- Multi-model orchestration: The system coordinates 19 to 20 different AI models — including specialized versions of Claude, Gemini, and Grok — routing subtasks to the most capable model for each step
- Remote control from anywhere: Users can assign objectives from any device; Personal Computer executes them on the Mac mini, with a full audit trail and a kill switch for immediate control
- Mac-only at launch: Available initially via waitlist to Perplexity Max subscribers at $200/month
Perplexity also launched Computer for Enterprise at the same conference, adding integrations with over 400 business applications — Salesforce, Snowflake, HubSpot, GitHub, Slack, Notion — and full enterprise compliance features including SOC 2 Type II certification and SAML single sign-on.
Why This Matters
For Mac Users Thinking About AI Productivity
The announcement crystallizes a trend that has been building since OpenClaw went viral earlier this year: the Mac mini is becoming the default AI agent hardware. Developers have been buying the $499 M4 Mac mini specifically to run local AI agent systems because it combines Apple Silicon performance with modest power consumption — roughly $3 to $5 per month in electricity costs when running around the clock. Perplexity's product is the first commercially polished version of an idea that was previously only accessible to engineers comfortable configuring their own agentic pipelines.
For Knowledge Workers With Complex Workflows
The appeal of a machine that keeps working while you are not at your desk is real. Perplexity ran an internal study measuring productivity gains: in four weeks, Computer completed what the company calculated as 3.25 years of equivalent work and saved $1.6 million in labor costs. Those numbers have drawn skepticism — a Hacker News commenter called it a “wild statement that does not seem to be supported by any actual data” — but the underlying proposition resonates. Long-horizon tasks like compiling research, processing inboxes, and generating reports are genuinely tedious and genuinely parallelizable by an AI agent with the right access.
What the Community Is Saying
Reaction has been mixed but leaning positive. Tech forums lit up with discussions about turning Mac minis into dedicated AI infrastructure, with many enthusiasts excited about the idea of a local agent that does not route sensitive data through a cloud browser session. The local-first framing — your files stay on your device, the agent runs on hardware you own — resonates strongly with privacy-conscious Mac users.
The skepticism is mostly about price and scope. $200 per month, plus the hardware investment, puts the first-year total at roughly $2,900 to $3,000. That is a meaningful commitment for a product still in waitlist access. Critics also point out that the open-source OpenClaw framework offers similar capabilities for the cost of API calls — typically $30 to $200 per month depending on usage — for anyone willing to do the setup work.
“The concept is genuinely exciting. A machine that just works on your behalf while you sleep — that's a real productivity unlock for the right person. The question is whether most people are the right person at $200/month.”
— Hacker News discussion thread, March 2026
What You Can Do Now
If You Are a Developer or Power User
Join the Personal Computer waitlist at perplexity.ai if the use case fits your workflow. A spare Mac mini at $499 running 24/7 as a personal AI agent is a compelling setup if you regularly have complex, multi-hour tasks — research compilation, data processing, inbox management — that could run unattended. If the $200/month subscription is steep, the OpenClaw framework on GitHub offers a DIY path to similar capabilities with more configuration work.
How This Fits Your Workflow
What Perplexity Personal Computer does well is long-horizon autonomous execution: give it a goal and it works toward it over hours or days. What it does not replace is the AI layer you interact with throughout your actual workday — the quick lookups, writing assistance, and document context that make moment-to-moment work faster.
For Mac users who want AI that works with their own knowledge — not just autonomously browsing the internet — tools like Elephas offer a complementary approach: a system-wide AI assistant at $9.99/month that builds a searchable Super Brain from your own documents, notes, and files. Where Personal Computer handles the tasks you hand off to a machine overnight, Elephas handles the AI you reach for a dozen times a day — with a ⌘+Space shortcut that works in any Mac application, grounded in your actual documents rather than the broader internet. Try Elephas free to see how it compares.
What's Next
Watch For
- Broader waitlist rollout: Perplexity has indicated Personal Computer will expand access through 2026. The current waitlist-only model is partly about managing infrastructure load for what is genuinely compute-intensive 24/7 AI orchestration.
- Windows support: Personal Computer is Mac-only at launch because the Mac mini's power efficiency and Apple Silicon performance make it an ideal host. Perplexity has signaled Windows support will follow.
- Apple's response: Apple is replacing Core ML with a new Core AI framework at WWDC 2026, with support for third-party model integration via the Model Context Protocol. That framework could make native macOS apps — not just dedicated agent products like Personal Computer — significantly more capable as AI coordinators. The Mac AI agent landscape is moving fast.
- Price compression: The $200/month price point will face pressure. OpenClaw remains free and configurable; as Apple's Core AI and on-device models improve, the cost of running capable local agents will decline.
Key takeaway: Perplexity Personal Computer is a meaningful milestone in the commercialization of local AI agents on Mac hardware. The $2,900 first-year cost puts it out of reach for most individuals today — but the concept of a Mac mini running AI on your behalf 24/7 is now a product, not just a developer experiment. The question for the rest of 2026 is how quickly the price and setup barrier drops to reach a mainstream audience.
Frequently Asked Questions
What is Perplexity Personal Computer?
Perplexity Personal Computer, announced March 11, 2026, is a software agent that runs continuously on a dedicated Mac mini. It connects your local files, apps, and sessions to Perplexity's AI infrastructure — acting as a 24/7 digital worker that handles tasks and coordinates workflows autonomously using up to 20 different AI models, including Claude, Gemini, and Grok.
How much does Perplexity Personal Computer cost?
Access requires a Perplexity Max subscription at $200 per month, plus a dedicated Mac mini (base M4 model starts at $499). The first-year total is approximately $2,900–$3,000 including hardware. Ongoing cost after that is $2,400/year for the subscription alone.
How does Perplexity Personal Computer differ from OpenAI Operator?
The key difference is scope. OpenAI Operator is a cloud browser agent — it only automates web-based tasks and cannot access local files. Perplexity Personal Computer runs on a physical Mac mini with full OS-level access to your local file system and applications. Both cost $200/month, but Personal Computer requires additional hardware and offers significantly deeper Mac integration.
Why are developers buying Mac minis for AI agents?
Apple Silicon Mac minis deliver strong AI compute performance with very low power consumption — roughly $3–5/month in electricity running 24/7. The $499 M4 base model is affordable as a dedicated machine. Local agents on Mac mini can access the full macOS file system, run continuously without disrupting a main work machine, and keep data on-device rather than in the cloud.