Waiting for Dot? Build your own AI assistant.

You want an assistant that helps move a project forward, not just one that answers a question and waits for another prompt. It should learn how you work, keep tasks moving and be reachable when you are away from your desk. That is what makes OpenAI's Dot interesting.

If you are waiting for personal access in Germany, OpenClaw and Miracles offer a way to start building that kind of collaboration today. Run OpenClaw on your computer or server, then take your conversations with you on iPhone and iPad using Miracles.

What is a Dot, and who can use one?

OpenAI describes dots as always-on agents in ChatGPT with their own cloud computer. They can work with connected apps, follow projects and take on tasks between conversations. You can also connect a personal computer.

For personal Pro accounts, OpenAI's help page currently limits access to adults outside the European Economic Area, Switzerland and the United Kingdom. That excludes Germany. Business Premium, Enterprise, Edu and Healthcare have separate access rules, so “Dot is unavailable throughout the EU” would be too broad. This is the position checked on 13 November 2026.

But account availability is only one question. Another is whether you want a managed assistant or would enjoy shaping your own.

A similar ambition, two different approaches

With Dot, OpenAI provides the managed environment. With OpenClaw, you decide where your agent runs, which supported model it uses and which tools it gets. That means more setup on your side, but also more room to make it yours.

OpenClaw is not another way to sign in to your ChatGPT dot. You are building a separate assistant. If you enjoy experimenting and making things your own, that is part of its appeal.

What an ordinary day could look like

In the morning, you discuss a small website idea with your agent at your computer. It creates a draft in the project folder. Later, you remember that the site needs to work well on phones. You open the same conversation in Miracles and add that thought.

With suitable development tools, your agent can keep editing and checking the draft. In the evening, you inspect the preview and decide what to do with it. Your ideas reach the place where the work is happening instead of being scattered across three note-taking apps.

For recurring work, you can set up an OpenClaw automation, such as a weekly summary of selected project files. The task needs to be saved, and the computer needs to be available at the scheduled time. “Do this next week” should result in a concrete schedule you can inspect.

Your computer, your setup, your choice of model

One useful part of running your own setup is deciding where project files and agent notes live and which tools can work with them. Start with one project folder and expand access when you have a reason.

The model is a separate choice. With a cloud provider, the content used for a request is sent to that provider. With a local model, that processing happens on your hardware. Search, browsers and connected apps may use external services too. Self-hosting describes who operates the environment; individual components determine where data goes.

Budget for model access and, if needed, hosting alongside Miracles. Ordinary chats and the basic connection do not require Miracles Pro. Your iPhone or iPad needs iOS or iPadOS 26 or later.

Start with one project, not twenty integrations

1. Choose a computer that stays on while you need it. An existing machine is a good first experiment if it meets OpenClaw's current requirements.
2. Set up OpenClaw and supported model access. Follow the official getting-started guide.
3. Create a small result on the computer: a learning plan saved as a file, for example. Ask where it was saved.
4. Connect Miracles. The setup guide covers gateway access, device approval and connections on the go.
5. Continue the same conversation over mobile data. Test the everyday situation you built the setup for.

Here is a first task for your configured agent:

I want to build a personal assistant on my own computer and use it on the go with Miracles.

First inspect my installed OpenClaw version, model access and available tools. Briefly explain what runs locally, what data goes to external services and what ongoing costs are involved. Use the matching current documentation for settings; do not request credentials in chat.

Help me choose one small everyday project and implement it in a separate test folder. Do not add more integrations yet. Show me the result and where it is saved. Do not publish anything or send messages to other people.

Then guide me through continuing the same conversation in Miracles over mobile data. If something is missing, identify the specific next step. Propose changes with their scope, a way back and a functional test, and wait for my selection before applying them.

You do not have to wait for the perfect setup

Dot makes the idea of a personal, working agent tangible for more people. OpenClaw gives you your own route if you want to shape the setup yourself. Miracles brings it into the part of your day that happens away from your desk.

The best starting point is a small project that genuinely helps. Find ideas in Your own AI assistant. To avoid some of the setup detours we took, read our OpenClaw setup tips.

Sources

• OpenAI: Introducing dots
• OpenAI: Getting started with your dot — access and requirements
• OpenClaw: Getting started
• Miracles: Setup