Portable Computer for Windows is now available on AMD Ryzen AI Max Series processors. It makes it easy to run local AI agents that work with your connected apps and local files. Kick off tasks or schedule recurring work that runs entirely on your device.

Portable Computer handles tasks locally, running the agent harness and model on your device, and asks your permission when it needs help from a cloud model.

Portable Computer works across connected apps like Slack, Gmail, and Google Drive, along with files on your device. Running tasks entirely locally keeps your files, data, and answers on device and doesn’t use Computer credits.

Portable Computer runs in the Perplexity app for Windows and is available to all Consumer and Enterprise Pro and Max subscribers. Get started:

@AMD Grateful to @AravSrinivas and the entire @perplexity_ai team for the partnership and for reimagining what the PC can be. The PC is becoming personal. Excited to see Perplexity Portable Computer bring that vision to life on @AMD Ryzen AI Halo. This is just the beginning 🚀

@AMD finally an agent that can read my messy downloads folder at 3am

@AMD AMD’s Ryzen AI Max is a beast for local agents. I love that it handles apps and files together without the cloud lag, my morning routine just got smoother watching you work through yours

@AMD Running agents totally locally means less reliance on sending everything to the cloud for simple stuff. That local control is the real hook here.

@AMD The recurring-work bit is the one I'd watch; local file permissions age quietly

@AMD 💯👇🏾

@AMD 本地代理完全在您的设备上运行，隐私和离线可用性都更稳

@AMD The permission step before calling a cloud model is a thoughtful detail. A clear view of which files a task can access would make this even easier for people to use with confidence.

@AMD local agents + recurring tasks could actually change how people use their pcs

@AMD Nice, running local agents on-device is a big win for privacy and latency. Curious how much of the Windows app ecosystem can actually take advantage of this yet. Excited to try it when I get a Ryzen AI rig.

Local agents solve an important part of the equation: execution close to the data. But once agents continuously work across local files and apps, the device also becomes a source of persistent agent state. The interesting challenge is making that history structured, traceable and reusable instead of letting each scheduled task start from scratch.

@AMD fix the context 400 error in the dgx spark

@AMD Local files staying local is the best part about this. Great update from both teams.

@AMD On-device only counts if Slack and Drive still write back without a cloud hop. A local harness that asks permission on the first miss is just a quieter API client.

@AMD Treat local agents like a capable assistant with the keys to your desk: start with a narrow folder and app scope, then expand access only after the workflow behaves reliably.

@AMD it asks before calling a cloud model. does it also ask before a scheduled task writes to slack or gmail?

@AMD Running agents against local files without sending everything to the cloud is a really useful direction.

@AMD Local AI agents running natively on hardware is the true milestone for privacy and autonomy. Great move. ⚡️

@AMD Local AI agents running directly on your device is a huge step forward.

@AMD qwen 3.8 27b is having a week. a ternary version built a full game on a 12gb 3060 and now it's powering local agents on AMD. 27b is turning into the local sweet spot

@AMD Running agents on the device, against your own files, is what I've been waiting for. Scheduling them to repeat locally is a nice touch.

@AMD Local agents are the moment AI stops being a tab and starts behaving like infrastructure. The winning UX is invisible execution with clear permissions and a useful log when something goes sideways.