Zoom AI Services: Introducing Summarizer and Translator APIs

We launched Zoom AI Services with the Scribe API as the first in a suite of enterprise-grade intelligence APIs. Today we're expanding that suite with two new APIs: Summarizer and Translator.

Together with Scribe, these APIs cover the full arc from raw audio to structured insight to localized output—without managing model infrastructure, storage pipelines, or queuing yourself.

Summarizer API: Derive intelligence from conversations

Transcripts are only useful when someone reads them. The Summarizer API does that automatically: send it conversational data from any system (VTT, SRT, or plain text) and it returns structured recaps, action items, and summaries in a single call or as a batch job.

Zoom brings deep expertise in conversational data to this workflow. The model is designed around how meetings and calls actually unfold: decisions, follow-ups, open questions, and commitments spread across speakers and time. It also gives developers predictable task-based outputs like recap, summary, and action items instead of relying on a general-purpose model whose response shape can change from prompt to prompt.

Batch mode processes up to 10,000 transcript files per job directly from your S3 bucket, writing one summary result per input file to your output location. Each result drops into your CRM, ticketing system, or user interface without additional processing.

The API supports four summarization tasks: recap, action_items, summary, and full_summary. The output can be returned in nine languages: English, Chinese, Japanese, Spanish, French, German, Portuguese, Italian, and Arabic. Use it to eliminate manual note-taking after sales calls, auto-populate support tickets, or run overnight summarization jobs across your full call archive.

Read our Summarizer blog or go straight to the Summarizer docs.

Translator API: Take your content global

Reaching a global audience means translating your docs, support articles, changelogs, and UI strings—and keeping them in sync every time something changes. The Translator API handles that at scale.

Batch mode accepts text files directly from your S3 bucket and translates them asynchronously, writing one output file per input to your destination. For individual strings, fast mode handles on-demand translation in a single synchronous call. Both modes support bidirectional translation between English and eight other languages: Chinese (Simplified), Japanese, Spanish, French, German, Portuguese, Italian, and Korean.

When you include a webhook URL in your batch job, the API delivers a signed notification on completion—so your publishing pipeline can deploy translated content immediately without polling.

Read our Translator API blog or go straight to the Translator API docs.

Get started

Zoom AI Services APIs are available on the Zoom Build platform. The ai-services-quickstart repository on GitHub includes a starter project to get you up and running with any of the APIs quickly. A natural first pipeline: run Scribe over a recording, pass the transcript to Summarizer for a recap and action items, then send those through Translator for every market you ship in.

We'd love to see what you build with these latest updates to Zoom AI Services.