01What Is Perplexity Computer Fully Local?

Perplexity Computer Fully Local is a new mode in the Perplexity AI agent that runs entirely on your Mac. It can open files, read notes, check mail, and complete tasks without sending any data to the internet. Your private client notes, business plans, and spreadsheets stay on your computer.

In this episode, Julian Goldie explains how this update, released on October 20, 2026, changes the game. Instead of mailing your documents to a cloud server, you get an AI assistant that works right on your desk. It uses Apple Silicon Macs and needs at least 36GB of unified memory to run the fully local version.

02Main Features of the New Local Mode

The new mode brings several powerful features that make it different from older cloud-based AI. Here are the key features Julian highlights:

  • Fully local operation – The AI model, planning, and tasks all happen on your Mac, not in a data center.
  • File access – It can open your notes app, check mail, and read spreadsheets in Numbers.
  • Multi-step tasks – It can plan and execute a sequence of actions, like a real assistant.
  • Two model options – You can use Qwen 3.6 or PPLX Pro, Perplexity's own trained version of Qwen, both running offline.
  • Privacy control – You decide when to use the cloud; private work stays local, and harder research can go online only if you allow it.

03How to Set Up Perplexity Computer Fully Local

Getting started with Perplexity Computer Fully Local is straightforward if you have the right hardware. Julian outlines the steps you need to follow:

  1. Check that you have an Apple Silicon Mac (like M1 or newer).
  2. Ensure your Mac has at least 36GB of unified memory for the fully local version.
  3. Open the Perplexity Computer app on your Mac.
  4. Select the Fully Local mode in the settings.
  5. Choose your model – either Qwen 3.6 or PPLX Pro.
  6. Grant the app permission to access your local folders and files.
  7. Start a task by typing a prompt that references your local files.

That’s it. Once you’re set up, you can begin handing off work to your local AI agent.

04How to Use It: Three Real Workflows

Julian demonstrates three practical workflows you can try right away. Each one uses your local files and keeps you in control.

Workflow 1: Welcome message for new leads – You prompt the agent to read your welcome notes folder and draft a friendly message. The result is a ready-to-send draft that mentions coaching calls and tutorials. You review and send it in under a minute.

Workflow 2: Content plan for social media – You ask the agent to look at your notes folder and build a five-day content plan. It returns a clean list with titles and key points, saving you hours.

Workflow 3: Member follow-ups – You provide a list of new members and ask for a personal check-in message for each. The agent drafts a stack of short messages, and you just read and send them.

In every case, you approve before anything goes out.

05Results and Comparisons from the Episode

Julian compares this new local mode to traditional cloud AI. He says cloud AI is like mailing your documents to a stranger; fully local AI is like hiring someone who sits at your desk and uses your own phone. That shift matters for privacy and comfort.

He also notes that local models may not be as smart as the strongest cloud models on hard reasoning tasks. But that’s not the point. The point is giving you a choice.

Private work stays private, and you can still use the cloud for bigger research tasks when you choose.

The episode shows real results: tasks that used to take an hour now take minutes, and you keep final control over everything.

06Tips, Limits, and Mistakes to Avoid

There are important limits to keep in mind. First, this only works on Apple Silicon Macs – not Windows or older Intel Macs. Second, you need at least 36GB of unified memory; otherwise, the fully local version won’t run.

Third, local models can struggle on the hardest reasoning tasks, so don’t expect them to match cloud giants on everything.

Julian also warns against letting the AI send anything without your approval. Always review drafts and plans before publishing. A common mistake is trying to use the cloud for everything, but this update is about giving you choice.

Use local for private work, and only go to the cloud when you need extra power.

07Who Should Use Perplexity Computer Fully Local?

If you’re a freelancer or consultant, this tool can take over tasks you hate doing by hand, like drafting proposals. If you run a small team, use it to clean up repetitive busy work so you can focus on client relationships. If you’re a business owner, start paying attention now – the gap between businesses that use this well and those that ignore it is growing fast.

Julian emphasizes that this tool doesn’t replace you; it removes the boring parts. You still do the thinking, decisions, and relationships. The time you save lets you focus on what matters.

He also mentions that most business owners haven’t noticed this update yet, so there’s a window of opportunity to get ahead.