Parallel tool calls: fan out Sume renders, wait once
ElevenAgents can run multiple tools in one turn. On Sume's hosted MCP, submit several jobs, then wait on all of them with one jobs_wait call of up to 20 ids.

The ElevenLabs changelog for Sep 21, 2026 says ElevenAgents can execute multiple tools in one turn when parallel tool execution is enabled. On Sume's hosted MCP the matching pattern is to submit several jobs, then call jobs_wait once with job_ids (1 to 20) and wait_for set to all or any.
Why one batch wait
Parallel tool calls cut the number of agent turns, but each paid create still needs its own idempotency_key. The slow part of media work is rendering, not the call, so the useful step is waiting on the whole batch at once instead of one id at a time.
What jobs_wait does with many ids
Sume's batch wait returns a status snapshot for every requested id.
| Item | Behavior |
|---|---|
| Ids per call | 1 to 20 in job_ids |
wait_for | all (default) or any; either way every id is reported |
| Slice length | Default 50 s, capped at 55 s on remote MCP |
include_results | Completed jobs come back with their results in results[] |
| Unknown ids | An unknown or foreign-workspace id fails the whole call |
Handling long renders
A wait that hits its slice returns wait_slice_expired. Call jobs_wait again with the same ids. Never resubmit the paid create, since the original jobs are still running and still bill.
If a render runs for minutes, repeat the wait rather than asking for a longer timeout; larger values are clamped and the response says so.
When to use script_run instead
For three or more independent calls of the same shape, the docs recommend script_run, which runs a short JavaScript program on the Sume side, can call tools in parallel, and returns the child jobs[] to wait on. Paid creates inside it still need their own idempotency_key.
Arguments for one batch wait over three submitted jobs:
{
"job_ids": ["job_a", "job_b", "job_c"],
"wait_for": "all",
"include_results": true
}Sources
Related posts
More in Developers
- Pause between narration lines: TTS pause markers or audio concat?
Deepgram Flux TTS allows pause markers of 500 to 3000 ms, max 8 per request. Sume's timeline audio concat joins lines with no gap. Where to put the pause.
- Performance Max text limits: 30, 90, 90 and 25 characters, counted
Performance Max headlines run 30 characters, long headlines 90, descriptions 90, business name 25. A short script counts generated copy before upload.
- 1080x1350 sent as aspect_ratio: what Ideogram, Grok, Imagen get
Send pixels instead of a ratio and Sume snaps to the nearest native ratio. Tested table for 1080x1350, 1200x628 and 1500x500 across four model families.
- Pre-flight tool access: tools_list on Sume's MCP
Notion added a tool that reveals connection-scoped capabilities before requests. Sume's MCP does the same job with tools_list, tools_schema and mcp_health.
Written by Sume