AI and media

AI agents that use Deepgram

UtilitySpaces gives an AI agent scoped access to Deepgram through its own encrypted connection, not your password. The agent can call 3 named Deepgram tools, and the 2 that write wait for a person's approval unless an admin has set the connection to run without asking. Every call is written to a hash-chained audit log with its arguments and its result.

What an agent can do in Deepgram

Transcribe audio from a URL and synthesize speech with Deepgram. An agent calls these by name, and matching is exact.

Every tool a Deepgram connection exposes, and whether it writes, which waits for approval by default
Tool	What it does	Waits for approval by default
List projects	List projects this key can access.	No — read only
Transcribe URL	Transcribe audio hosted at a public URL (billed call) using the account's default model. Returns JSON transcripts. Feature options (model, language, diarize, ...) are query-string-only on Deepgram and cannot be set through this connector.	Yes — it writes
Text to speech	Synthesize speech from text (billed call) with the account's default voice; the voice/model is a query-string option on Deepgram and cannot be set here. Returns raw audio bytes, not JSON: the body arrives unparsed, so use it to confirm synthesis rather than to fetch a usable audio file.	Yes — it writes

1 read tool, 2 write tools. Generated from the connector catalogue the product runs on, so it cannot list a tool that does not exist.

What teams hand an agent in Deepgram

Teams with audio to process hand an agent both directions of Deepgram: Transcribe URL turns hosted audio into JSON transcripts, Text to speech synthesizes a reply, and List projects confirms what the key reaches. Both conversions are billed writes, so each waits for approval unless an admin has set this connection to run writes without asking.

Transcribe a recorded call or voice note from its URL and post the transcript into the thread (billed; after a person approves it)
Turn a short script into speech with the account's default voice (billed; after a person approves it)
Confirm which projects the key can access before wiring it into a workflow.

Before you let it write, or keep out: Model, language, and diarization are query-string options on Deepgram's side and cannot be set through this connector -- runs use the account's defaults. Text to speech returns raw audio bytes, so treat it as a synthesis trigger, not a file fetch.

What waits for a person

Reads run on their own. These do not — each one waits in Approvals for an owner or admin unless an admin has set this connection to run writes without asking, and every call toDeepgram, approved, rejected or unattended, is logged.

Transcribe URL POST
Text to speech POST

What connecting Deepgram takes

Deepgram uses an API key you create in Deepgram and paste in once. A workspace admin sets the connection up; the fields are:

API Key

Credentials are encrypted and bound to your workspace, never read back, and every outbound call passes one guarded fetch — the connector security posture sets out exactly how, and the security page covers the platform around it.

Where Deepgram fits

Give an agent Deepgram work by writing a message in the inbox, or put its tools in a scheduled flow. What the run cost is on the cost breakdown.

Other ai and media integrations: AnythingLLM, AssemblyAI, D-ID. Or see every application agents can connect to.

See it doing your actual work

Open a demo workspace, connect one app, and give an agent a real job. No card, no sales call.

Try the demo See pricing

We reply to every message within one business day.