Watcher agent

​

Overview

The Watcher agent is a Lancey feature that assigns an AI agent to monitor your tasks and provide intelligent nudges to keep them progressing, just like a human would. This allows you to work on more things simultaneously while your agent ensures tasks move forward efficiently.

​

How it works

Once a Watcher agent is assigned to a task, it continuously:

• Monitors progress - Tracks task execution and identifies blockers
• Detects issues - Identifies when a task gets stuck or encounters problems
• Provides nudges - Suggests next steps or adjustments to keep momentum
• Maintains context - Keeps track of task history and requirements
• Adapts dynamically - Adjusts approach based on task progress and results

This creates a more human-like workflow where an agent actively helps move work forward rather than just executing a single planned path.

​

Key features

​

Managing Watcher agent

​

Viewing assignment status

When you open a task, the Properties panel on the right side of the screen shows whether a Watcher agent is currently assigned to the task.

​

Unassigning Watcher agent

If you want to disable the Watcher agent for a specific task:

​

Reassigning Watcher agent

To re-enable Watcher agent on a task that has been unassigned:

1. Open the task
2. Look in the Properties panel for an Assign Watcher agent option
3. Click to reassign a new agent to monitor the task

​

When to use Watcher agent

Use Watcher agent in these scenarios:

• Complex tasks - Keep complex, multi-step tasks on track automatically
• Low-priority work - Let agents handle routine tasks while you focus on important decisions
• High volume - Manage more tasks simultaneously with agent assistance
• Uncertain requirements - Agent can adapt when requirements shift during execution

​

When to unassign Watcher agent

Consider unassigning Watcher agent when:

• You need full manual control over task execution
• The task is simple and doesn’t need monitoring
• You’re running experiments that require precise, non-adaptive behavior
• You want to review and approve each step before the agent proceeds
• You’re working with sensitive or high-risk changes

​

Best practices

Keep agents assigned for routine work

Let Watcher agents handle standard fixes and improvements. This frees your team to focus on strategic decisions and complex problems.

Monitor high-risk tasks manually

For critical or sensitive tasks, consider unassigning the Watcher agent to maintain tighter control and require explicit approvals at each step.

Use in combination with Auto Processing

Pair Watcher agent with Auto Processing for a fully autonomous workflow that monitors and executes tasks with minimal human intervention.

Review agent decisions

Check the task history to see how Watcher agent guided the task. Use this insight to refine your auto-processing rules and task instructions.

Create team policies

Establish team guidelines about when Watcher agents should be assigned or unassigned based on task type, complexity, and risk level.

​

Troubleshooting

Watcher agent isn't responding to blockers?

Verify that:

• The Watcher agent is still assigned in the Properties panel
• Your task instructions are clear and specific
• The agent has sufficient context about the issue
• Your AI model configuration is set up correctly

Task seems stuck with Watcher agent assigned?

Try:

• Checking the task history to see what the agent attempted
• Reviewing error messages or logs for specific blockers
• Providing additional context or clarification to the task
• Manually unassigning and reassigning a fresh agent

Can I assign a different Watcher agent?

When you unassign a Watcher agent, a new agent will be automatically assigned on next task update. All Watcher agents have the same capabilities, so you’re not selecting between different types.

How do I know if Watcher agent made changes?

Review the task history and activity log to see:

• Actions the agent suggested
• Changes it made or attempted
• Course corrections it provided
• Any errors it encountered

​

Next steps