Demodesk
ProductivityConnect AI agents to Demodesk, the sales meeting and coaching platform. Agents list recordings, retrieve transcripts, summaries, and coaching scorecards, and react when new recordings land so meeting insights flow straight into follow-up workflows.
What This Integration Enables
It is worth being precise about what this connector is, because Demodesk is often described by its scheduling. The public API v2 exposes users and recordings. There are no demo scheduling or lifecycle endpoints here, and demo events are delivered through webhooks configured in the Demodesk interface rather than programmatically, which is why the connector offers a polling trigger on new recordings instead of a real-time one. So treat this as a meeting-intelligence connector: recordings, transcripts, AI summaries, and coaching scorecards.
That is a strong hand for an agent. FlowRunner agents can list recordings with server-side filters on status, post-processing status, access level, host, attendee, demo ID, and creation date, fetch a single recording's participants and engagement statistics, retrieve a transcript as structured JSON or merged plain text with optional on-the-fly translation, pull transcripts for up to a hundred recordings in one call, and read the summaries and coaching scorecards attached to a call. Post-processing status done is the signal that transcript, summaries, and scorecards are all ready, and a well-built flow waits for it rather than polling blind.
The output is conversation content: what a customer actually said, and what a manager privately wrote about how a rep handled it. Moving that around is easy, which is exactly why the human-in-the-loop design on this connector is about where the content goes rather than about approving a write.
Without FlowRunner
With FlowRunner
Use Case Scenarios
Meeting summary onto the opportunity
On New Recording fires with a post-processing status of done. The agent calls List Recording Summaries, takes the summary generated from the configured prompt, and writes it onto the matching opportunity in HubSpot along with the participant list and the recording link from Get Recording. The rep's CRM notes are written before they have finished their next call, and they reflect the conversation rather than a recollection of it.
Quarterly call analysis
An enablement team wants to know how a new objection is being handled across the whole team. The agent calls List Recordings filtered by created date range and post-processing status, then calls Batch Get Transcripts for up to a hundred recordings at a time, walking the cursor until the set is complete. It loads the text into Snowflake for analysis with the recording tokens preserved so any finding can be traced back to the call it came from. Access filters are applied at the query, not after the export.
Follow-up commitments that do not get dropped
Customers leave calls with promises attached: a security questionnaire, a pricing approval, a reference call. The agent reads the transcript with Get Recording Transcript, extracts the commitments made by the seller, and opens tasks in Salesloft with the timestamp of the sentence that created each one. Where the commitment is ambiguous, the agent does not invent a due date. It surfaces the sentence and lets the rep decide whether it was a commitment at all.
Human-in-Loop Highlight
The sensitive artifact on this connector is not the transcript, it is the scorecard. List Recording Scorecards returns the template, the overall score, the reviewer and reviewee user IDs, an optional free-text comment, and per-question detail with criteria, scores, and comments. That is a manager's candid written assessment of a named person, written for a coaching conversation between two people. And the blast radius is set by whose key the agent holds: a Demodesk API key inherits the permissions of the user who created it, so a key created by a company admin can read every scorecard in the organization.
An agent building a weekly coaching digest is one template variable away from broadcasting a written comment about a named rep to an audience that was never part of that coaching relationship. So the agent checks its own reach and then stops. It calls Get Current User, reads the role back, and when the role is manager or company admin it treats scorecard comments as content it may read but not redistribute. It posts to the reviewer: "Coaching digest for this week includes 3 scorecards you submitted. Two carry written comments. Share as written and attributed, share scores and criteria only, or exclude these from the digest?" Scores that the team has already agreed are public metrics keep flowing. The sentence a manager wrote for one person moves only where that manager said it could.
Agent Capabilities
8 actionsUsers
2- Get Current User Returns the Demodesk user that owns the configured API key, with name, email, role of user, manager, or company admin, locale, time zone, and group memberships. Used to verify credentials and, just as usefully, to let a flow understand how much it can see.
- List Users Lists active users visible to the key, with case-insensitive partial search across full name and email and cursor pagination. Used to resolve a host or reviewer to a person.
Recordings
6- List Recordings Lists recordings with server-side filtering by status, post-processing status, access level, host, attendee, demo ID, and creation date range, with cursor pagination. Host and attendee accept an email address or the literal value me. A post-processing status of done means transcript, summaries, and scorecards are ready.
- Get Recording Fetches full detail for one recording by its public token, including host, meeting location, participants, engagement statistics, status, and links.
- Get Recording Transcript Retrieves a transcript as structured JSON with per-sentence timestamps or as merged plain text, with optional on-the-fly translation into another language. Available once post-processing is done.
- Batch Get Transcripts Retrieves transcripts for up to a hundred recordings in one call, keyed by recording token, with a per-recording status of ready, processing, not requested, empty, or not found. Used for bulk export and analysis without a hundred separate calls.
- List Recording Summaries Lists the AI-generated summaries for a recording, each with its prompt, language, and both plain-text and HTML content. Multiple summaries appear when several summary prompts are configured.
- List Recording Scorecards Lists the coaching scorecards submitted for a recording, with template, overall score, reviewer and reviewee IDs, optional comment, and per-question criteria, scores, and comments. Read it to route coaching feedback, and treat the written comments as belonging to the reviewer.
Triggers
1 triggersEvent Triggers
1- On New Recording Fires when a new meeting recording appears on the account, emitting the recording object. It can be scoped to a lifecycle status such as Ready, or required to wait for a post-processing status such as Done, which is the state where transcript, summaries, and scorecards are all available. The first poll establishes a baseline without firing, so recordings that already existed are not replayed into a flow, and the polling interval is configurable down to thirty seconds.
Frequently Asked Questions
What can FlowRunner do with Demodesk?
FlowRunner agents can run Get Current User, List Users, and List Recordings in Demodesk, plus 5 more actions.
Does connecting Demodesk to FlowRunner require OAuth?
No. Demodesk connects to FlowRunner with an API key, no OAuth flow required.
Can Demodesk trigger a FlowRunner workflow automatically?
Yes. Demodesk supports 1 trigger that can start a FlowRunner workflow automatically.
Start building with Demodesk
$100 in credits. No card required. Connect in minutes.