Understanding Background Agents in Vox

Most AI assistants stop the moment you look away. Vox has background agents that keep working while you do other things — research, document generation, multi-step workflows, all running autonomously. Here's how they work.

What are background agents?

Background agents are long-running tasks that Vox executes asynchronously via its task engine. When a task would take more than a few seconds — like researching a topic, generating a comprehensive report, or processing hundreds of files — Vox spawns an agent that works independently.

Spawning a task

The agent starts immediately and you can continue with other work. The task runs locally on your Mac via Electron scheduling.

Checking on tasks

Journal and checkpoints

Every background agent maintains a journal — a running log of what it's done, what it's found, and what decisions it's made. This serves two purposes:

• Rollback — If an agent gets stuck or takes a wrong path, it can roll back to the last checkpoint and try a different approach
• Transparency — You can review exactly what the agent did and how it arrived at its conclusions

Agents learn from experience

After completing a task, agents store trigger-solution patterns. If you ask for a similar research report next month, the agent already knows:

• Which sources were most reliable last time
• What format you preferred for the output
• Which comparison frameworks worked best
• How detailed you wanted the analysis

Best practices

• Be specific — “Research X and compare Y using Z criteria” gives better results than “look into X”
• Specify output — Tell the agent whether you want a file, an email, or just a verbal summary
• Check in periodically — Ask for status updates, especially on complex tasks
• Chain results — Use task output as input for other commands (“send that report to the team”)

Background agents are what make Vox more than a chatbot. They're autonomous workers that operate on your behalf — and they get better with every task.