Agents

Agents are the brains behind Duster's automation. They are autonomous entities capable of observing conversations, understanding context, and taking action to help your product team move faster.

What is an Agent?

An Agent is an AI-driven worker that orchestrates Workflows in response to Signals. Unlike a simple chatbot, a Duster Agent is integrated deep into your product development process. It doesn't just chat; it works.

• Observer: It monitors channels for relevant feedback, bugs, or feature requests.
• Analyst: It classifies messages, determines priority, and extracts structured data.
• Doer: It can create Linear issues, post replies, or alert specific team members.

Core Components

To understand how an Agent works, it helps to break it down into its three main components:

1. Signals: The events that wake the Agent up. This could be a new message in a Slack channel, a comment on a ticket, or a system event.
2. Workflows: The "playbook" the Agent follows. A workflow is a sequence of logic—deciding what steps to take based on the input.
3. Tools & Inputs: The capabilities the Agent has at its disposal (e.g., "Create Issue", "Reply to Thread") and the data it uses to make decisions.

Agent Persona: "Dusty" 🧑‍✈️

By default, your agent acts as Dusty, a helpful and precise Product Manager assistant.

• Tone: Professional, concise, and helpful.
• Goal: To clarify ambiguity and ensure every piece of feedback lands in the right place.

You'll see Dusty in action when it replies to threads to confirm bug reports or asks for missing details.