Some practical ways to get started

A short checklist for your first two weeks, with links to go deeper. Start soon, but start small.

Getting started checklist

1. Pick one collaboration space, three to five people, and one agent.
   Start with a small team that has a project to complete together. Teams with high trust work best for initial experimentation.
2. Configure a multiplayer agent.
   There are lots of ways to do this, the most common being Claude Tag ↗ or the Claude Managed Agents API ↗.
3. Make all the ways the team works open and shared by default.
   If a decision happens in a meeting or a DM, someone posts it the same day. For an agent, if it isn't written down and accessible, it doesn't exist. Why working in public matters ↗
4. Write the roster, agents included.
   List every recurring job and give each job one owner. Ask the agent to propose the list from what it can read, then let people decide. Defined roles with the right tools ↗
5. Write down the north star and pin it where the agent can read it.
   A north star is an ambitious goal that's measurable. People set the north star and agents help work towards it.
6. Give the agent one visible job first.
   A morning briefing shared to the entire team is a great task to start with. For the first few days, a person should review the briefing and give the agent feedback on its effectiveness. How Slack's teams started ↗
7. Build checks that agents can use.
   For every task you ask an agent to do, give it a rubric for what "passing" looks like. This scales trust the most quickly between humans and agents.
8. Extend trust a step at a time, per kind of task.
   After the agent has handled one kind of task well several times in a row, widen what it may do on that task. Build trust over time ↗
9. Teach the agent your way of working.
   When you explain the same thing twice, turn it into written instructions for the agent. Tutorial: teach Claude your way of working using skills ↗
10. Use a weekly retro to improve your process.
    Have agents and humans share what's working and what requires adjustment at the end of every week. Ask the agents on your team to save these so other agents can use them later.

Go deeper

• Building effective human-agent teams ↗
  The post this course is built on: the four practices and the five readiness questions.
• Turning conversation into knowledge: how Slack builds human-agent teams ↗
  How one company's teams started small and grew trust.
• Academy course: AI Capabilities and Limitations ↗
  A working picture of what the models behind these agents can and cannot do.
• Academy course: AI Fluency: Framework & Foundations ↗
  The habits of working well with AI that every teammate brings to the shared room.
• Academy course: Introduction to agent skills ↗
  For engineering teams: how written skills let an agent reuse your team's ways of working.
• Docs: Claude Code in Slack ↗
  For engineering teams: an agent that works from your team's Slack channels.
• Docs: Agent Skills overview ↗
  The technical reference behind skills, for whoever sets the agent up.
• Academy tutorial: What is Claude Managed Agents? ↗
  For whoever will build your team's agent: a four-minute look at how Claude can run an agent with its own tools and memory.

What's next

Congratulations, you've explored what it takes to build a healthy human-agent team. Most of these practices were healthy team habits long before agents, but agents make them costlier to skip. Check your learning in the next lesson with a short quiz.