Phone Computer: will workspace habits become consumer habits?

The hypothesis

Workspace habits will become consumer habits over time. Consumers will sometimes want an assistant specific to the task in front of them, and people with different personas will still have the same job to be done.

The shorthand is “instinct meets Dropbox.” A familiar place for your files, with an assistant you can ask to do something with them. The bet is about how consumers get things done, not just access to a computer from a phone.

Phone Computer is the new bet: a personal cloud computer with an assistant you text. Put chosen work in a folder on your Mac, text what you need, and get the result back in the folder. No new phone app.

The question is whether habits developed around assistants at work carry into personal life. Will consumers bring their own files, delegate a specific task, review the result, and come back when that job comes up again?

A persona describes the person. The job describes the progress they want to make. The experiment should look for shared jobs across different people, rather than assume each persona needs a separate product.

Why this bet follows the last one

The consulting bet exposed a slow path to a buying decision: referrals, then the time it takes for someone to recognize the problem they need help with. AI moves faster than that cycle.

Phone Computer puts a concrete interaction in front of someone: here are your files, here is a task, here is the result. The bet is that trying it can reveal value faster than explaining a consulting engagement. That still needs evidence.

The product loop

1. Put the files you want help with in a chosen folder on your Mac.
2. Text a task from your phone.
3. Let the assistant work on those files in a private cloud runtime.
4. Review the proposed result and approve changes.
5. Find the approved result back in the folder.

The folder is the boundary. The product should work only on what the user chooses to share. Asking for work should not silently authorize publication or overwrite newer edits.

Current evidence

The project has a development implementation and a landing page. Its hosted workflow has been exercised locally with a demo assistant and local runtime. That is implementation evidence, not proof of customer demand or a production-ready service.

The live integrations and the native Mac experience still need verification. No customer traction, repeat usage, or willingness to pay is claimed here.

Proposed experiment

Find a recurring job involving personal files that consumers with different personas want done. Learn how they handle it today and whether they already use an assistant for a similar task at work.

Then observe the whole Phone Computer loop: setup, the text request, review, and finding the result back in the folder. Look for the same underlying job across participants, even when their circumstances and language differ.

Record whether the result was useful, how much intervention it needed, and whether the person chooses to use it again. A waitlist signup is interest; a second useful task is stronger evidence.

What would change the decision

Evidence for continuing would be consumers across different personas returning for the same underlying job, carrying an assistant habit from work into personal use, understanding what the assistant can access, and using the reviewed results.

Evidence against it would be workplace assistant habits failing to transfer into personal use, no shared recurring job across the consumers tested, setup friction that outweighs the convenience, tasks that routinely need the user back at the Mac, or results that take more effort to fix than to make.

The bet ends on December 31, 2026. At that point, review the evidence and decide whether to continue, change direction, or stop.

The target pilot group, investment cap, and numerical success criteria remain to be set. This entry records the direction and a proposed test, not an agreed experiment budget.