Skip to main content
Dispatches several RunAsync calls simultaneously on one Agent instance. Each run gets its own Temporal workflow (or local goroutine) — no separate Agent per request required. Source: examples/agent_with_concurrent_runs/

​
What it demonstrates

  • RunAsync + sync.WaitGroup for fan-out across one agent
  • Results printed as they arrive (arrival order varies)
  • Runtime-agnostic: local and temporal both supported via AGENT_RUNTIME

​
Run

From examples/: 
# local runtime (default)
go run ./agent_with_concurrent_runs

# pass your own prompts
go run ./agent_with_concurrent_runs "Who wrote Hamlet?" "What is sqrt(144)?" "Name a primary color."

# Temporal runtime
task infra:temporal:up && task infra:temporal:wait
AGENT_RUNTIME=temporal go run ./agent_with_concurrent_runs

​
Key code

a, _ := agent.NewAgent(
    agent.WithName("concurrent-agent"),
    agent.WithLLMClient(llmClient),
    // ...same options for local or Temporal
)

var wg sync.WaitGroup
resultCh := make(chan result, len(prompts))

for i, prompt := range prompts {
    wg.Add(1)
    go func(idx int, p string) {
        defer wg.Done()
        asyncCh, _ := a.RunAsync(context.Background(), p, nil)
        ar := <-asyncCh
        resultCh <- result{idx: idx, prompt: p, res: ar.Result, err: ar.Error}
    }(i, prompt)
}

go func() { wg.Wait(); close(resultCh) }()

for r := range resultCh {
    fmt.Printf("Q: %s\nA: %s\n\n", r.prompt, r.res.Content)
}
Each RunAsync returns immediately; the goroutine blocks on <-asyncCh until that run completes. All runs are in flight simultaneously on the same Agent.

​
Expected output

Q: Who wrote Hamlet?
A: Hamlet was written by William Shakespeare.

Q: What is sqrt(144)?
A: The square root of 144 is 12.

Q: Name a primary color.
A: Red is a primary color.
Arrival order varies — results print as each run finishes, not in submission order.

​
Learn more

Multiple Agents

Separate agent instances with distinct task queues

Run Async

Non-blocking RunAsync with result channel