TikTok's Next Episode: a brand series as one Format and one queue

If you plan a brand series after TikTok's The Next Episode, produce it as one Sume Format (the recipe that fixes the look, voice and structure) plus one bulk queue per season. A queue takes 1 to 100 ordinary Format runs and keeps 1 to 16 of them in flight, so a ten-episode season is one request instead of ten scripts running on your laptop.

The program itself is reported, not documented by Sume. Metricool's TikTok news page (read 2026-10-03) says TikTok launched The Next Episode with Amplify to help fund creator-led series. That page does not say who is eligible or how much funding applies, so check TikTok's own program terms before you plan around money. This post covers only the production side.

What a series needs from a pipeline

A series repeats the same decisions every week: same character, same framing, same length, new story beat. In Sume that maps to two layers. The Format holds the recipe, a SKILL.md plus reference files, authored in the Agents dashboard, in chat, or over the Contents API. The call holds what changes: your instruction and a JSON input object. The Formats overview calls this the how and the what, and the recipe is established before your instruction, so you are not re-sending a system prompt on every episode.

A season queue

The envelope is two keys, concurrency and items. Each item is the same body you would send to a single run, so it must name at least one of instruction, input, previous_run_id or attachments. The 202 receipt already shows the first concurrency items running.

curl -sS -X POST "https://api.sume.com/v1/formats/acme/brand-series/bulk-runs" \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: brand-series:s01:v1" \
  -d '{
    "concurrency": 2,
    "items": [
      { "input": { "episode": 1, "beat": "The delivery goes wrong" },
        "generation_spend_cap_usd": 10 },
      { "input": { "episode": 2, "beat": "The neighbour knows" },
        "generation_spend_cap_usd": 10 },
      { "input": { "episode": 3, "beat": "The reveal" },
        "generation_spend_cap_usd": 10 }
    ]
  }'

What the queue does and does not do

• Each queue item is a normal Format run: one sandbox, one agent turn, one receipt. A queue is not a different engine.
• The queue has no webhook. communication.webhook_url is per item, and you poll GET /v1/format-run-queues/{id} for progress.
• Queue status completed means every item is terminal, not that every item succeeded. Read counts.failed and counts.canceled.
• A bad item fails the whole create with 400 invalid_request and details.index, before any queue exists. A child that fails after the 202 fails only its own slot.
• Sume does not post to TikTok, apply to Amplify, or manage a program application. It produces the files; you publish them.

Where to go next

Send a fresh Idempotency-Key per season so a retry of the same request never starts a second one; see the Bulk runs page for replay behaviour. If episode 4 needs a reshoot, continue that run with previous_run_id instead of paying for the whole queue again, as described in Runs and results. For a character that must look the same each week, start from the Formats guidance in Calling a Format.