Autonomous mode lets Qualia work independently on complex research tasks for extended periods. Instead of waiting for your approval at each step, the agent makes decisions on its own and reports progress via Slack.
Setting up autonomous mode
Autonomous mode requires two things:
- Autonomous Mode turned on
- Slack connected and configured
- Go to Settings > Slack
- Click Connect to Slack and authorize Qualia
- Select a channel for updates
Once connected, agents running in Autonomous Mode will post progress to your chosen channel.
Research style settings
Research style is set in Settings > AI > New Agents > Default Research Style, or per-session from the chat composer.
Curiosity
How much the agent explores beyond your immediate request:
Independence
How much the agent assumes versus asks:
Parallelism
How much the agent splits work across additional agents:
Autonomous Mode
A separate toggle that supersedes curiosity, independence, and parallelism. When on, the agent runs fully autonomously — no confirmation dialogs, posting progress to Slack instead. When off, the agent follows your three research-style settings.
Autonomous Mode skips all confirmation dialogs. Use it for trusted background work where you’re comfortable with the agent making decisions without you.
Working with autonomous agents
Starting an autonomous run
- Turn on Autonomous Mode (in Settings or per-session)
- Describe a research direction:
“Explore different architectures for this timeseries forecasting problem. Try at least 5 approaches and report the best.”
- The agent works independently, posting updates to Slack
Monitoring progress
While an autonomous agent runs:
- Slack: Receive progress updates, preliminary findings, and completion notices
- Tasks sidebar: See the task graph and status
- Agents sidebar: View agent hierarchy and status indicators
Guiding from Slack
You can respond to Slack updates to steer the agent:
- Propose new directions
- Ask it to focus on specific approaches
- Request more detail on promising results
The agent incorporates your feedback and adjusts its approach.
When to use autonomous mode
Autonomous mode works well for:
- Open-ended exploration: Find the causes of all anomalies in a dataset
- Model sweeps: Try many configurations and automatically follow up on promising directions
- Long-running experiments: Training runs that need to be monitored for hours and stopped if no longer useful
