Motive - Maximize parallelization and autonomy with Astra on your VM

Motive is a wiki for agent collaboration that runs on your VM. It supports Codex, Claude Code and open source models. How to install

Runs on your VM

Motive is built for people who run their agents on a VM. While you get all the benefits of an always on machine with the internet access of a data center, the agentic interface tends to be somewhat hacky. People are using Termius and the Remote section of the ChatGPT app has been a big leap forward. But we think there is still plenty of room for a new interface. If you're new to running your agents on a virtual machine, I recommend Hetzner for their cheap prices and reliable service.

A wiki to see what's happening and collaborate with your agents

Motive stores your project pages and local workspace files on your VM. Relevant context is sent to your configured model provider, and connected services receive the data needed to carry out your requests. You and your agents can work with the local files directly.

Bring your own AI

Motive builds on your existing Codex CLI or Claude Code setup. We are rehearsing the Linux installation flow and offering assisted early trials. The public demo lets you explore the interface; it does not run agents against your own workspace. Installation time and end-to-end setup on a fresh machine are still being verified.

Tag your agent in comments

Instead of just interacting via turn-by-turn conversation, you can ask your agent to produce a document and then tag your agent in comments. Comments handle context automatically - forking from the original conversation that generated the document and working through comments in the doc one by one so that your shared context on the doc and the underlying work compounds.

Listen and comment on the go

With audio mode, you can continue collaborating while out on a walk. Tap the audio symbol and you get an enjoyable ElevenLabs readout of any doc so you can consume it on the go. Leave a comment using your voice at any point and it's automatically transcribed and added in writing. CarPlay support coming soon.

Computer use

The progress in computer use over the last months has been amazing and led OpenAI to call out the AGI era with the release of GPT-6. When your agent now needs to use a computer (e.g. use a website that doesn't have an API or use software without an API), it can just do so on your VM, and you see what it does via a PIP screen on your chat. When the agent needs help from you (a CAPTCHA it won't solve by itself for compliance reasons, logging you into a website), it hands off to you. Support for copy + paste of passwords and other information in and out of the machine is built in. Computer use works on both desktop and mobile.

Embed apps in your wiki

Pages your agent has built to interface better with you (e.g. dashboards, 3D visualizations, presentations) can be embedded where they belong. Create a hub for your project, link its key dashboard right above the place where your agent writes its business updates.

Websites can be embedded as standalone pages or interactive elements inside another page.

Scheduled prompts and agent triggers

You and your agent can add a block to a page to schedule any prompt to run at a future point. Type / and select Scheduled prompt to have it run at a specific time. Type / and select Trigger to have an agent or code run when something specific happens (e.g. you receive an email or Slack message from a given sender or someone signs up on your website).

Long-horizon agents

To give an agent an ongoing project, type / and select Agent. Enter its goal and choose how often it should run; the default is every 12 hours. The first run starts when you commit the goal. Later runs receive the latest project page and recent results, with links to earlier runs. You can review the work and steer it with comments attached to the relevant part of the page.

Collaborate on mobile

The Motive mobile app lets you do anything you can on mobile, too. Be notified when it's stuck, review your wiki or respond to your agents.

Coming soon: Shared spaces

Collaborate with others in a shared workspace. You can share individual pages or share an entire team space where you and your agents can review and collaborate on docs, media and apps. Each person defines the way they want their agent to collaborate with those of others in the workspace.

How to install

On your VM, open an interactive terminal or connect over SSH as the Linux account that will own Motive. Supported systems are Ubuntu 22.04, 24.04 or 26.04 and Debian 12 or 13, with systemd, on x64 or arm64. A direct root login is supported; run the command without a sudo prefix.

curl -fsSL https://motiveapps.ai/install.sh | bash

Choose Claude or Codex when prompted. Setup installs the required software, reuses compatible existing CLI installations and logins, and guides you through any required provider or Tailscale sign-in. System package installation may ask for your sudo password. Your workspace stays private on your own Tailscale network.

Keep the terminal open until setup says “Motive is ready” and prints your private workspace link. Setup checks the website, provider connection and iPhone update download before showing that link. Open it and start your first conversation in Motive; you do not need to ask an agent to run the installation. If setup is waiting for a signed repair, leave the terminal open so it can continue.