Openbase

About Openbase

Openbase is an open-source tool for managing a team of AI coding agents through voice commands from a phone or other device, without requiring a screen or desk. You can dispatch tasks, steer agents mid-work, and approve sensitive changes by speaking. The software is licensed under AGPL-3.0 and is currently in a private beta with a waitlist.

Review

Openbase addresses a specific friction point in AI-assisted development: supervising coding agents still tends to pull you back to a laptop, even though the agents themselves work asynchronously. The tool shifts supervision to a voice interface that you can access while away from your desk. It syncs with your machine, so you can pick up where the voice session left off when you sit down again.

Key Features

• Voice-based dispatch and steering. Speak a task to an agent, adjust its direction while it works, and approve or reject sensitive actions out loud - no screen interaction needed.
• Approval checkpoints. Agents pause before executing potentially destructive actions like commits, pushes, or file changes. They explain what they intend to do and wait for an explicit verbal confirmation ("yes proceed") before continuing.
• Threads view. Displays the agent's full transcript and groups related code diffs across multiple files, so you can review changes visually when you return to your machine.
• Skills system. Lets you save reusable instructions that agents follow across sessions - for example, always summarizing a task back before starting long-running work.
• Local approval queue. Pending approval requests are stored on the host machine. If your phone drops connectivity while an agent waits for a decision, the request reappears in the Approvals tab once you reconnect.

Pricing and Value

Openbase is open source under the AGPL-3.0 license. The hosted version includes a free trial period, after which it costs $20 per month for voice and text-to-speech model usage. There is also a developer pathway: you can clone the GitHub repository and run a setup script that configures free local models like Kokoro, bypassing the monthly fee. Access currently requires joining the waitlist during the private beta phase.

Pros

• Voice commands let you dispatch tasks and approve changes without sitting at a laptop.
• Approval checkpoints require explicit verbal confirmation before sensitive actions execute, and each approval gets logged with the action and outcome.
• Threads view groups related diffs for multi-file change review alongside the agent's transcript.
• The system fails closed on garbled or partial transcriptions, re-asking rather than guessing on approvals.
• Open-source codebase with a self-hosted option means you aren't locked into the hosted plan.

Cons

• Reviewing complex multi-file diffs by audio alone remains challenging; the tool is still improving how much context gets summarized without a screen.
• Currently in private beta with a waitlist, so access is not immediate.
• Not well suited for developers who work in consistently noisy environments where background sound may disrupt voice interaction, or in workplaces that restrict phone use during shifts.

Who Openbase Works For

Openbase fits developers who already use AI coding agents and want to monitor or steer them during stretches away from a desk - on a walk, during a commute, or in other mobile contexts. The approval checkpoint design and audit logging make it usable beyond experimental repositories, though teams handling highly complex merges may still prefer a screen-based fallback for detailed diff review. Developers curious about voice-driven agent supervision can join the waitlist and test it as the beta expands.

Open 'Openbase' Website