Agents & Departments
Meet your AI team: who they are, how they're organized, and what they can do for you.
Leo runs a project management tool. His marketing department has a Content Writer drafting blog posts, an SEO Analyst picking keywords, and a Social Manager scheduling posts. His sales department has a Lead Researcher finding prospects and an Outreach Specialist drafting cold emails. Leo reviews the work each morning, approves what looks good, and sends back anything that needs changes. His 3-person team just got 5 extra hands.
What Can You Ask Your Agents?#
Agents work best with clear, specific instructions. Here are real examples organized by department:
Marketing#
- "Write a blog post about [topic]"
- "Research SEO keywords for our landing page"
- "Draft 5 social media posts for this week"
Operations#
- "Summarize this week's completed tasks"
- "Build a report on our team's output this month"
- "Create a project timeline for the product launch"
Sales#
- "Research 10 leads in the fintech space"
- "Draft a cold outreach email for [company]"
- "Update our CRM with notes from today's calls"
Support#
- "Draft a reply to this customer complaint"
- "Update the FAQ with the new refund policy"
- "Analyze support ticket trends from last month"
Engineering#
- "Summarize the latest PR changes"
- "Write changelog entries for v2.3"
- "Triage these 5 bug reports by severity"
How Agents Work#
Each agent is an AI worker with a specific role: a Content Writer, a Lead Researcher, a Bug Triager. You describe what you need in plain language and the right agent picks it up. Agents can work independently or wait for your approval before delivering results.
You can customize each agent's name, department, skills, which tools it can access, and whether it works autonomously or checks in with you first.
Departments#
Agents are organized into departments that mirror a real company. During onboarding, departments are set up based on your business type, but you can reorganize anytime.
| Department | Example Agents | Typical Tasks |
|---|---|---|
| Marketing | Content Writer, SEO Analyst, Social Manager | Blog posts, keyword research, social scheduling |
| Operations | Project Coordinator, Data Analyst, Report Builder | Task routing, dashboards, weekly reports |
| Sales | Lead Researcher, Outreach Specialist, CRM Manager | Prospect lists, email sequences, pipeline updates |
| Support | Ticket Handler, Knowledge Writer, Feedback Analyst | Reply drafts, FAQ updates, sentiment analysis |
| Engineering | Code Reviewer, Doc Writer, Bug Triager | PR summaries, changelog entries, issue triage |
Supervised vs Autonomous#
This is the most important setting on each agent.
Supervised agents pause after completing work and send the result to your approval queue. Nothing goes out until you review it. This is the default.
Autonomous agents complete their work without waiting. Great for low-risk, high-volume tasks where you trust the output.
You control this per agent. Keep Support supervised while letting Operations run autonomously. Change it anytime from the agent detail page.
Agent Statuses#
Each agent shows one of four statuses: Idle (ready for work), Working (on a task), Waiting Approval (finished, needs your review), or Paused (manually paused by you).
Adding & Removing Agents#
Add new agents from the Agents page. Choose a role template or create a custom agent with your own configuration.
Removing an agent archives it. Task history is preserved, but it stops accepting new work. You can restore archived agents anytime.
Tip