Skip to content

Agent Teams

Agent Teams let you group agents together so they can collaborate, delegate work, and communicate with each other. Instead of building one agent that tries to do everything, you create specialists and organise them into teams with clear communication rules.

A team is a named group of agents that can interact with each other according to rules you define. Teams model the way departments work in a real business — a customer support team might have a triage agent, a technical specialist, and an escalation manager, all working together to resolve issues.

When agents are on a team:

  • They can send messages to each other
  • They can delegate tasks to teammates
  • They can be discovered by other agents (using the find_teammates tool) when their membership is public
  • Communication between members follows the rules you set

Teams can also include people. The User Members tab lets you add human team members alongside agents — useful when a workflow needs a person to approve or review the team’s work.

  1. Navigate to Agent Teams

    Go to the Agent Teams section in your Sprigr Teams portal at team.sprigr.com.

  2. Click “New Team”

    Click the New Team button to open the Create Agent Team form.

  3. Enter a name and description

    Give the team a clear name like “Customer Support”, “Operations”, or “Sales”. Add a description that explains the team’s purpose — this helps other agents (and your team members) understand what the group does. Click Create Team.

  4. Add member agents

    The new team opens on its detail page. From the Members tab, add the agents you want to include. When adding each member you choose its visibility — Public or Private (see below). You can always add or remove members later.

  5. Set communication rules (optional)

    From the Communication Rules tab, add any allow, deny, or route-through rules the team needs. Without rules, team members can communicate freely.

Communication rules control how agents within a team are allowed to talk to each other. There are three rule types:

Explicitly permit messages between two agents. Note that adding any allow rule switches the team into allowlist mode: from then on, only explicitly allowed communications are permitted. Use this when you want tight control over who talks to whom.

Block two agents from communicating with each other. Use this to enforce boundaries. For example, you might deny direct communication between a sales agent and a finance agent to ensure all financial requests go through proper channels.

Messages between two agents must pass through a designated intermediary agent. This is useful for:

  • Approval workflows — All requests to the procurement agent must go through the operations manager agent first.
  • Quality control — Customer-facing messages are reviewed by a supervisor agent before being sent.
  • Logging and compliance — An intermediary agent can log all communications between two departments.

Each team member has a visibility setting that controls whether agents outside the team can discover it through the team:

  • Public — Agents across your organisation can discover this member and send it messages (subject to communication rules). Use this for members that serve the whole organisation, like customer support or IT help desk.
  • Private — Only members of the team can see this agent’s membership. Use this for sensitive roles like HR, finance, or executive operations.

You can toggle a member between public and private at any time from the Members tab, and set the team’s default visibility for new members in the team’s settings.

Every team can optionally designate an orchestrator — a coordinator agent that oversees the team’s work. The orchestrator has special capabilities:

  • CC on all messages — With the orchestrator CC setting enabled, the orchestrator is copied on every message exchanged between team members, giving it full visibility into the team’s activity.
  • Task routing — Other agents can send requests to the team, and the orchestrator decides which member should handle it.
  • Coordination — The orchestrator can assign tasks, check progress, and ensure nothing falls through the cracks.

The team’s settings also let you set a performance tier override for the orchestrator, so coordination work can run on a more capable tier than the agent’s default.

Customer support escalation

A triage agent handles initial enquiries. If the issue is technical, it delegates to a technical specialist on the same team. If the customer is unhappy, it escalates to a supervisor agent. An orchestrator tracks all open issues.

Department routing

A front-desk agent receives all incoming requests and routes them to the correct department team — maintenance, accounts, or scheduling. Each department team has its own specialists and communication rules.

Multi-step workflows

A quote request comes in. The sales agent gathers requirements, the estimating agent calculates costs, and the approvals agent checks margins. The team orchestrator ensures each step completes before the next begins.

Knowledge specialists

A research team has agents specialising in different areas — compliance, market data, and competitor analysis. When a question comes in, the orchestrator routes it to the right specialist.

  • Creating Agents — Build the specialist agents you need before adding them to a team.
  • Agent Settings — Configure each agent’s persona, tools, and integrations.
  • Companion Agents — Learn how your personal companion agent can delegate tasks to team members on your behalf.
  • Tools Reference — See the send_message, delegate_task, find_teammates, and manage_team tools that enable agent-to-agent collaboration.