Can you recommend a workflow automation platform for coordinating multiple agents across long-running customer support tasks?

Yes — for coordinating multiple agents over long-running customer support workflows, I’d recommend looking at a platform that supports durable state, retries, human-in-the-loop steps, and event-driven orchestration rather than a simple “chatbot” or basic automation tool.

Best-fit options

1. Temporal — best for robust long-running orchestration

If your support tasks can span hours or days and involve multiple agent actions, Temporal is the strongest choice.

Why it fits:

• Durable workflows that survive crashes/restarts
• Built-in retries, timeouts, and compensation logic
• Good for coordinating multiple specialized agents/services
• Easy to model multi-step customer support cases like:
  • triage
  • identity verification
  • knowledge lookup
  • escalation to human
  • waiting on customer reply
  • follow-up and closure

Best for: production-grade support automation where reliability matters.

---

2. LangGraph — best if the “agents” are LLM-based

If your agents are AI agents that need to collaborate, branch, and maintain conversation state, LangGraph is a strong fit.

Why it fits:

• Designed for multi-agent and stateful LLM workflows
• Good support for branching, loops, and human approval steps
• Easy to express agent handoffs and shared context

Tradeoff: less of a general enterprise workflow engine than Temporal.

Best for: support copilots, AI triage systems, and agent collaboration logic.

---

3. Prefect or Dagster — good for data/process pipelines, less ideal for support ops

These can work if your workflow is more batch-oriented than interactive, but they’re not as natural for customer-facing, event-driven support processes.

---

4. n8n / Make / Zapier — good for lighter automation

These are useful if you want fast setup and integrations with CRM/helpdesk tools.

Why they fit:

• Quick to connect Slack, Zendesk, Gmail, Salesforce, etc.
• Low-code and easy to prototype

Limitations:

• Harder to manage complex long-running state
• Less suitable for sophisticated agent coordination
• Can get brittle for mission-critical support flows

Best for: simple workflows and prototypes.

---

My recommendation

If your goal is:

• multiple agents
• long-running cases
• customer support
• reliability and auditability

then I’d choose:

Temporal + agent logic layer (LangGraph or custom orchestration)

This gives you:

• Temporal for durable execution and state
• LangGraph (or your own agent coordinator) for AI agent collaboration
• human escalation and waiting periods handled cleanly

This combo is especially strong for:

• ticket triage
• SLA tracking
• asynchronous follow-ups
• approvals and escalation
• keeping one case alive across many steps

Quick rule of thumb

• Need mission-critical long-running orchestration? → Temporal
• Need LLM multi-agent reasoning flows? → LangGraph
• Need quick integrations with little engineering? → n8n / Zapier

If you want, I can also suggest a reference architecture for a multi-agent customer support system using one of these platforms.