Zoom AI · Agentic Search · APIs

Zoom Search API vs Ask API: when to use which?

By Michael Obraitis, Managing Partner, and Michael Dale, CEO · Published August 28, 2026 · Last reviewed August 2026 · ADAM Pulse Knowledge Base

Verification date.

Product names, APIs, connector lists, licensing notes, and MCP setup steps were checked against Zoom and Microsoft documentation on 28 August 2026. These details change. Confirm against live vendor pages and your own tenant before you buy, connect, or process regulated data.

Short answer

Use Search API when your agent should see the evidence and your model (or your policy layer) should decide what to do with it. Use Ask API when the agent should ask a question and receive a grounded answer, with Zoom running the retrieval loop so transcripts and fan-out never enter the agent's context window.

Same context layer. Different contract. Search API is “here are ranked hits.” Ask API is “here is the answer Zoom synthesized, with citations.” Zoom's licensing footnote: Search API needs an eligible Zoom Workplace or ZoomMate paid plan; Ask API needs ZoomMate plus available Zoom AI credits.

Architecture and deployment sit in the complete guide. MCP wiring sits in Connect Claude to Zoom MCP.

How do the two APIs differ?

Search API versus Ask API as Zoom documented them on 11 August 2026.
	Search API	Ask API
Job	Ranked retrieval for a bring-your-own model	End-to-end retrieval plus synthesis
What lands in the agent context window	Ranked results (still much smaller than brute-force opening every meeting, in Zoom's test)	One synthesized, cited answer
Who ranks and re-asks	Your agent can loop; the layer still does ranked search	Zoom's server-side agentic retrieval loop
Best fit	Custom agents, audit of retrieved chunks, your own summarizer or policy model	Worker Q&A, low-latency answers, clients that should not see raw transcripts
Zoom-stated plan need	Eligible Zoom Workplace or ZoomMate paid plan	ZoomMate paid plan and available Zoom AI credits
Token claim to qualify	60–80% fewer tokens versus brute-force in Zoom's Claude / 89-meeting week test	Described as fewest tokens; numeric Ask figure is a projection pending a direct benchmark

When is Search API the right default?

You need to show citations in your own UI and let a reviewer open the source.
You already run a governed model (including on-prem or a specific subprocessor) and will not send synthesis to Zoom's Ask path.
You are building evaluation: you want the retrieved set so you can score recall, not only the final sentence.
Policy must run on retrieved text (DLP, allow-lists, “never quote this repository”) before any answer is produced.

Zoom's Claude test is the illustration: asking the layer beats listing and opening 11 meetings. That is a Search API result. It is still retrieval, not a finished briefing, unless your model writes one.

When is Ask API the right default?

The user wants an answer, not a result list.
You do not want transcripts in a third-party agent's context window.
Fan-out across Zoom plus CRM plus files should happen once, server-side, instead of your agent learning each connector.
You have ZoomMate seats and AI credits and accept Zoom's synthesis quality, with human review for anything consequential.

Ask API is not “Search API but cheaper, guaranteed.” Zoom has not published a completed Ask-versus-Search token benchmark; it published a projection. Credits also meter Ask. Budget the credits in the pilot, not after launch.

What mistakes should you avoid?

Calling Ask API and then also stuffing Search hits into the prompt. You pay twice and you put the transcripts back in the window Ask API was meant to keep out.
Treating either API as write access to Salesforce or ServiceNow. Knowledge data sources are retrieval. Action connectors are a different ZoomMate catalog. Do not conflate them.
Skipping identity mapping and then blaming the API for missing files.
Quoting 60–80% in a business case as a committed saving. Cite Zoom's methodology paragraph or omit the number.

Frequently asked questions

What is Zoom Search API?

The evidence mode of Agentic Search. It returns ranked results from the unified context layer so your model can reason over them. Zoom says it requires an eligible Zoom Workplace or ZoomMate paid plan.

What is Zoom Ask API?

The answer mode. Retrieval and synthesis run server-side. The agent receives one grounded, cited answer. Zoom says it requires a ZoomMate paid plan and available Zoom AI credits. Zoom's published Ask API token figure is a projection pending a direct benchmark.

Which one uses fewer tokens?

Zoom's 11 August 2026 post says Ask API uses the fewest tokens of the paths it compared, because intermediate search stays off the agent's context window. The 60 to 80 percent reduction it headlines is the Search API versus brute-force test, not a measured Ask API guarantee. Qualify both figures.

Can I use both?

Yes. Use Search API when you need to inspect or re-rank evidence, apply your own policy model, or keep a transcript of retrieved chunks. Use Ask API when the product is a short grounded answer and you accept Zoom's synthesis. Many pilots start with Search API for observability, then add Ask API for worker Q&A.

Are these the same as Zoom REST APIs?

No. REST covers meeting CRUD, users, phone, and similar resources. Agentic Search APIs are retrieval over the context layer. Zoom's MCP product servers also expose tools such as search_meetings and search_zoom that are semantic retrieval, not REST filters. Confirm the live tools/list on the server you connected.

Where do I call them?

Zoom says both are delivered over MCP to Claude, Codex, custom UIs, and other MCP-capable agents. Use the endpoint and scopes from the current Zoom MCP server documentation for your app. Zoom remote MCP does not support Dynamic Client Registration.

This page is part of the ADAM Pulse Zoom Agentic Search cluster. Existing Zoom AI Companion guides are related, not duplicates.

References

Primary vendor documentation. Where a vendor has not published a claim, this article says so rather than filling the gap.

ADAM Pulse can help design an Agentic Search pilot

ADAM Pulse helps Zoom customers pick sources, map identities, choose Search API versus Ask API, connect Claude or other MCP clients, and run a phased pilot with human review. This is a design engagement, not a product claim about Zoom. Support: (888) 989-4872 · support@adampulse.us

Prepared by ADAM Pulse (USA Telecom Consulting LLC) — managed Zoom and network services. Support: (888) 989-4872 · support@adampulse.us