Attention

A read-only API key integration that syncs Attention sales call transcripts into Rippit on a schedule. For what applies to every integration, see how data integrations work. For the security model, see Data integration security.

Connection summary


Authorization	Attention organization-level API key, sent as `Authorization: Bearer <key>`
Key level	Organization-level only. A user-level key, or a key for an organization with no teams, is refused when you connect
Scope parameter	None. An organization-level key reads every team’s calls
Writes to Attention	None. Every call is a `GET`
Credentials shared with Rippit	The API key. Rippit stores it and uses it for every sync
Credential check	When you connect, Rippit reads your organization from the key, then tests the key against Attention before the connection is saved. A rejected key saves nothing
Connection identity	Your Attention organization UUID, read from the key. There is nothing to name
Required Rippit role	Workspace admin
API base	`https://api.attention.tech`
API version	`v2`, in the path
Sync method	Scheduled polling
Audio and video	Not downloaded. The stored call record includes a signed video playback token that Rippit never uses. See Data handling
Maximum historical backfill	90 days
Field selection limit	150 columns

Before you connect

Rippit asks for one value, an API key. An Attention admin creates it under Settings → Organization → API Keys. Create an organization-level key: Rippit identifies the connection by the organization the key belongs to, and a user-level key cannot report one. An organization-level key reads every team’s calls in your Attention organization.

Data handling

Question	Answer
Is call content ingested, or only metadata?	Content, as text: the full transcript with speaker names, plus the call record as Attention returns it, minus the word-level transcript. The record includes title, owner, participants, team, duration, whether the call was internal, Attention’s scorecard results (scorecard titles, summaries, and items), CRM link fields, and the call’s link in Attention. The recording itself is not ingested
Are audio or video files stored?	No. Rippit downloads no audio, video, or other files from Attention. Duration is read as a number from the call record
Does Rippit keep any access to the recording?	The stored call record includes `videoUID`, a short-lived, signed video playback token, and the meeting’s Zoom join link when the call has one. Both are stored as Attention returns them. Rippit never uses either to fetch media or join a meeting
Is the raw transcript stored as Attention returns it?	No. Attention returns the transcript word by word. Rippit merges words into speaker turns and stores the turns in place of the word list
Are calls without a transcript ingested?	No. Calls Attention marks pending or failed are filtered out of every request, and a call returned with no transcript words is skipped
Can Rippit modify Attention?	No. No write code path exists

What a call becomes in Rippit

Each Attention call becomes one conversation in Rippit.

Rippit record	Content
Subject	The call title
Agent	The call owner, by email: the Attention user the call belongs to, or, when there is no user, the organizer participant’s email if Attention provides one. The owner is also added to your agent list
Fields	Owner name, participant emails when Attention provides them, team name, duration, whether the call was internal, and the call’s link in Attention (`web_url`, `https://app.attention.tech/conversations/<id>`). Other single-value attributes on the call record are also available to select. Attributes that hold a list are stored with the record but are not offered as fields
Default selected fields	Title, owner, participant emails, and team name
Transcript comments	The transcript, one comment per speaker turn, each timestamped at the call’s start plus the turn’s offset

Consecutive transcript segments from one speaker join the same turn until there is a pause of more than 3 seconds or the turn reaches 1,000 characters. Each turn carries the speaker name Attention identified, or the speaker’s email when there is no name. A turn with neither is attributed to a generic participant. A call’s start is the time its live stream started, or its creation time when Attention reports no stream start. The owner’s display name is the Attention user’s first and last name, or the organizer participant’s name. Speaker and owner emails are stored in lowercase.

API endpoints

Purpose	Endpoint	Method
Organization lookup when you connect	`/v2/organizations/teams`	GET
Credential check when you connect, and before each sync	`/v2/conversations/list?size=1&filter[hide_transcript]=true`	GET
Call discovery and content	`/v2/conversations/list?fromDateTime=…&toDateTime=…&page=…&size=10&detailedTranscript=true&filter[hide_pending]=true&filter[hide_failed]=true`	GET
Field discovery, a sample of 5 calls, only when Rippit has no stored sample	`/v2/conversations/list?page=1&size=5&detailedTranscript=true&filter[hide_pending]=true&filter[hide_failed]=true`	GET

The organization lookup reads the organization UUID from the first team the key returns. It is why a key that cannot list teams cannot connect. There is no separate per-call fetch. detailedTranscript=true returns each call’s full transcript inline in the list, so one paged request both discovers calls and reads their content. No other Attention object is listed or swept. Owner, participants, and team come from the call record itself.

Sync behavior


Call discovery	Windowed on the call’s creation time (`fromDateTime` to `toDateTime`), 10 calls per page, page-number pagination
Window width	7 days
Trailing buffer	Each window closes 1 hour behind the current time
Window size limit	A window that pages past page 1,000 (10,000 calls) stops the sync with an error rather than skipping calls
Daily re-sweep	The last 24 hours are listed again once a day, because Attention lists a call only after its transcript finishes processing
Update handling	Calls are selected by creation time, not by last edit. A change made in Attention to a call already synced is picked up only while the call falls inside the daily re-sweep
Expected latency	A call syncs once it is at least 1 hour old and Attention has finished processing its transcript. One that finishes later arrives with the daily re-sweep

Rate limiting

Every request, retries included, starts at least 1 second after the previous one, so a sync sends at most 60 requests per minute.
Each page of up to 10 calls is one request. Each sync also spends one request on the credential check.

Getting Started

Connecting your Data

Using Rippit

Connecting to your AI Agent

Use cases

Reference

Connection summary

Before you connect

Data handling

What a call becomes in Rippit

API endpoints

Sync behavior

Rate limiting

​Connection summary

​Before you connect

​Data handling

​What a call becomes in Rippit

​API endpoints

​Sync behavior

​Rate limiting

Connection summary

Before you connect

Data handling

What a call becomes in Rippit

API endpoints

Sync behavior

Rate limiting