Genesys Cloud - Main

Good morning, Hive Mind!

I'm fairly sure this has come up before, but I wanted an update to see if anyone has a solution.

I need to be able to record audio from within an Architect flow (Inbound or In-Call), Specific use cases (right now):

1. When requesting a callback, the requirement is to prompt the caller for their name and have the agent listen to that name before placing the call. Proposed solution(s) so far:
   1. Use a Voicebot and perform STT - client doesn't want to use AI at the current time.
   2. Use a customized voicemail - Caller would lose their place in queue.
2. Client wants to give supervisors the ability to update prompts via their telephone. (For things like emergency notices.)

Has anyone managed to crack this one? It seems to be a fairly common ask.

TIA

#ArchitectandDesign
#Telephony

------------------------------
Paul Simpson
Views expressed are my own and do not necessarily reflect those of my employer.
------------------------------

------------------------------
Att,
Breno Canyggia Ferreira Marreco
------------------------------

Thanks,

That is what I feared. I had hoped Genesys would've fixed this by now - it's not an uncommon requirement!

------------------------------
Paul Simpson
Views expressed are my own and do not necessarily reflect those of my employer.
------------------------------

Original Message:
Sent: 06-08-2026 11:29
From: Breno Canyggia Ferreira Marreco
Subject: Recording Audio from a Flow

Hi Paul, This is not currently supported natively in Genesys Cloud CX Architect

This type of use case typically requires an external integration (e.g., AudioHook or a custom recording service).

------------------------------
Att,
Breno Canyggia Ferreira Marreco
------------------------------

Hi Paul,

My understanding is that Architect does not currently provide a native way to record an arbitrary audio clip during a flow and then make that recording available for later playback.

For the callback use case, we've taken a slightly different approach. Rather than recording the caller's name, we identify the customer using information collected during the interaction. For example, this could be through a CRM lookup using a customer identifier entered by the caller, or by matching the ANI/phone number where appropriate. The customer details can then be presented to the agent within the agent script when the callback is delivered, which may help address the underlying requirement without needing to store and replay an audio recording.

For the prompt management use case, we've had success using Data Tables to manage announcement content. This allows authorized users to update messaging without republishing the Architect flow. While it doesn't provide telephone-based recording, it does remove the need to modify the flow itself when announcements need to change.

That said, if the requirement is specifically to capture and store audio for later playback, I'm not aware of a native Architect capability that supports that today.

Hope this helps.

------------------------------
Phaneendra
Technical Solutions Consultant
------------------------------

Hey Paul,

I looked at the Product Ideas Lab and I didn't see anything similar to what you are trying to do in your original post. However I did find this idea for your second major bullet point about emergency notices being updated via a phone call. It is currently marked as Future Consideration in the Product Ideas Lab. I would recommend submitting an idea for your other points, and commenting and voting on the other idea.

 I would recommend submitting s

------------------------------
Jason Kleitz
Online Community Manager/Moderator
------------------------------

I think this Idea refers to his original post.

------------------------------
Levy Mertz
Communications Manager
------------------------------

Original Message:
Sent: 06-09-2026 14:48
From: Jason Kleitz
Subject: Recording Audio from a Flow

Hey Paul,

I looked at the Product Ideas Lab and I didn't see anything similar to what you are trying to do in your original post. However I did find this idea for your second major bullet point about emergency notices being updated via a phone call. It is currently marked as Future Consideration in the Product Ideas Lab. I would recommend submitting an idea for your other points, and commenting and voting on the other idea.

 I would recommend submitting s

------------------------------
Jason Kleitz
Online Community Manager/Moderator
------------------------------

HI Paul,

You mentioned on your post BOT option...

Had a customer asking same thing, wanted ability for caller to leave a message, I used a BOT to capture the message and used the transcription from that and presented in the agent script page.

It is a transcript rather than audio clip but have found, in general, the quality of transcription is very accurate.

This of course is a slight variation to ask, and requires additional BOT and charges for use. I think its 17min per token and 15sec slots on transcript.. still an additional cost though.

Simon

------------------------------
Simon Brown
Maintel Europe Limited
Senior Applications Consultant
------------------------------

Hey Paul, the recording use case is certainly interesting.

I'm going to start with the "easier one", number two. I'd recommend using  Voice bot to collect an utterance for the prompt change, then allow the person to hear a TTS playback of the utterance and if they approve then write that utterance to a field in a data table. That then is consumed by the flow. You could create a menu structure for various different data table fields, things like inclement weather versus emergency closure versus holiday closure, and then they would choose which one they want to update, and then you'd have a purge process that would be able to either turn off the data table field or delete the utterance. And by delete the utterance I mean replace with blank audio or something like that. To be honest though, I'd actually consider doing this as like a web messaging deployment on a publicly accessible site. Where the manager, supervisor of the call center would interact with a chatbot and then just type in the prompt that they want. Then doing all the steps above, inserting that prompt into whatever data table field is necessary.

As for creating an ad hoc recording and storing just a name of the recording and attaching it to the callback I would believe that is not going to be natively possible. At least without using Voicemail. You could reprioritize the voicemail in the queue to not lose their spot, but that would require semi-complex logic to accurately age the voicemail based on all the current calls time in queue.

------------------------------
Steve Alix
EDCi
------------------------------

Hi Paul,

For #1, I believe it might be possible to develop a custom solution that would handle the use case.

To start with you need to record the caller's name.  You can do that today from Architect by making a Genesys Cloud Data Action request to call this API to start/stop the recording of a snippet of audio: https://developer.genesys.cloud/devapps/api-explorer#post-api-v2-conversations-calls--conversationId--participants--participantId--snippet-record.  That recording snippet will be a separate audio file saved alongside the entire recording in Genesys Cloud.

To play that audio file back to the agent at the start of handling the callback you would need to develop a small web app that would authenticate with Genesys Cloud using a PKCE grant OAuth client, use the Recording APIs (https://developer.genesys.cloud/devapps/api-explorer#get-api-v2-conversations--conversationId--recordings and https://developer.genesys.cloud/devapps/api-explorer#get-api-v2-conversations--conversationId--recordings--recordingId-) to retrieve the specific snippet recording, retrieve the audio file using the signed S3 URL provided by the Recording APIs, and finally use native Javascript to play the audio to the agent.  It sounds complicated but it is really straightforward to develop.  This web app can be embedded in an agent script using a Web Control and will run automatically when the agent script displays, or could be controlled via a custom button action.

If you would need help with any of this, Genesys Professional Services could consult & assist to help you with the development or they can build it for you.  If that is of interest to you then please reach out to your Genesys account manager.

------------------------------
Jim Crespino
Senior Director, Developer Evangelism
Genesys
https://developer.genesys.com
------------------------------

Thanks, Jim.

A couple of things.

1. The documentation (personal bug-bear of mine) doesn't give details of the returned objects from API calls (It's very good at providing the status codes though!). I'm assuming the URL of the recording is in there, but what is it called / how is it structured? - Normally, I would execute the action from the explorer and look-see, but that's a little hard to do with this one!
2. Is ParticipantID (I'm assuming for the caller?) exposed as a built-in variable, or will it require another Data action to get it?
3. I see you mention using a second Data Action to stop the recording. Is there any way to detect when someone stops speaking, or perhaps timeout if they don't say anything?
4. I know it wasn't part of the initial ask, but I can see others wanting to play the recording back to the caller for confirmation. Can we do that too? Another DA, or using a Play Audio?

I appreciate your assistance!

------------------------------
Paul Simpson
Views expressed are my own and do not necessarily reflect those of my employer.
------------------------------

Original Message:
Sent: 06-11-2026 10:30
From: Jim Crespino
Subject: Recording Audio from a Flow

Hi Paul,

For #1, I believe it might be possible to develop a custom solution that would handle the use case.

To start with you need to record the caller's name.  You can do that today from Architect by making a Genesys Cloud Data Action request to call this API to start/stop the recording of a snippet of audio: https://developer.genesys.cloud/devapps/api-explorer#post-api-v2-conversations-calls--conversationId--participants--participantId--snippet-record.  That recording snippet will be a separate audio file saved alongside the entire recording in Genesys Cloud.

To play that audio file back to the agent at the start of handling the callback you would need to develop a small web app that would authenticate with Genesys Cloud using a PKCE grant OAuth client, use the Recording APIs (https://developer.genesys.cloud/devapps/api-explorer#get-api-v2-conversations--conversationId--recordings and https://developer.genesys.cloud/devapps/api-explorer#get-api-v2-conversations--conversationId--recordings--recordingId-) to retrieve the specific snippet recording, retrieve the audio file using the signed S3 URL provided by the Recording APIs, and finally use native Javascript to play the audio to the agent.  It sounds complicated but it is really straightforward to develop.  This web app can be embedded in an agent script using a Web Control and will run automatically when the agent script displays, or could be controlled via a custom button action.

If you would need help with any of this, Genesys Professional Services could consult & assist to help you with the development or they can build it for you.  If that is of interest to you then please reach out to your Genesys account manager.

------------------------------
Jim Crespino
Senior Director, Developer Evangelism
Genesys
https://developer.genesys.com
------------------------------

Hi @Jim Crespino!

Sorry, to hassle you on this one, but if you could look at my queries above, I'd appreciate it!

------------------------------
Paul Simpson
Views expressed are my own and do not necessarily reflect those of my employer.
------------------------------

Original Message:
Sent: 06-22-2026 13:36
From: Paul Simpson
Subject: Recording Audio from a Flow

Thanks, Jim.

A couple of things.

1. The documentation (personal bug-bear of mine) doesn't give details of the returned objects from API calls (It's very good at providing the status codes though!). I'm assuming the URL of the recording is in there, but what is it called / how is it structured? - Normally, I would execute the action from the explorer and look-see, but that's a little hard to do with this one!
2. Is ParticipantID (I'm assuming for the caller?) exposed as a built-in variable, or will it require another Data action to get it?
3. I see you mention using a second Data Action to stop the recording. Is there any way to detect when someone stops speaking, or perhaps timeout if they don't say anything?
4. I know it wasn't part of the initial ask, but I can see others wanting to play the recording back to the caller for confirmation. Can we do that too? Another DA, or using a Play Audio?

I appreciate your assistance!

------------------------------
Paul Simpson
Views expressed are my own and do not necessarily reflect those of my employer.
------------------------------


Original Message:
Sent: 06-11-2026 10:30
From: Jim Crespino
Subject: Recording Audio from a Flow

Hi Paul,

For #1, I believe it might be possible to develop a custom solution that would handle the use case.

To start with you need to record the caller's name.  You can do that today from Architect by making a Genesys Cloud Data Action request to call this API to start/stop the recording of a snippet of audio: https://developer.genesys.cloud/devapps/api-explorer#post-api-v2-conversations-calls--conversationId--participants--participantId--snippet-record.  That recording snippet will be a separate audio file saved alongside the entire recording in Genesys Cloud.

To play that audio file back to the agent at the start of handling the callback you would need to develop a small web app that would authenticate with Genesys Cloud using a PKCE grant OAuth client, use the Recording APIs (https://developer.genesys.cloud/devapps/api-explorer#get-api-v2-conversations--conversationId--recordings and https://developer.genesys.cloud/devapps/api-explorer#get-api-v2-conversations--conversationId--recordings--recordingId-) to retrieve the specific snippet recording, retrieve the audio file using the signed S3 URL provided by the Recording APIs, and finally use native Javascript to play the audio to the agent.  It sounds complicated but it is really straightforward to develop.  This web app can be embedded in an agent script using a Web Control and will run automatically when the agent script displays, or could be controlled via a custom button action.

If you would need help with any of this, Genesys Professional Services could consult & assist to help you with the development or they can build it for you.  If that is of interest to you then please reach out to your Genesys account manager.

------------------------------
Jim Crespino
Senior Director, Developer Evangelism
Genesys
https://developer.genesys.com
------------------------------

Hi Paul,

Sorry for ghosting you on this one.

I've passed along your comment about the API documentation to the product owner of the Developer Center.  Most of the APIs do describe the response payload/format if you expand the 200 OK response, but it is true that some APIs don't do a great of documenting the response payload.

For #2, Participant ID is not exposed in Flows or Scripts, so you'll need to make a Data Action request to obtain that id for the specific participant.

For #3,  I'm not aware of a way to monitor if/when there is a pause in the customer speaking.

For #4, there is no way that I'm aware of to play any kind of audio that is not a pre-defined prompt file.

I hope that answers your questions.

 

------------------------------
Jim Crespino
Senior Director, Developer Evangelism
Genesys
https://developer.genesys.com
------------------------------

Original Message:
Sent: 06-22-2026 13:36
From: Paul Simpson
Subject: Recording Audio from a Flow

Thanks, Jim.

A couple of things.

1. The documentation (personal bug-bear of mine) doesn't give details of the returned objects from API calls (It's very good at providing the status codes though!). I'm assuming the URL of the recording is in there, but what is it called / how is it structured? - Normally, I would execute the action from the explorer and look-see, but that's a little hard to do with this one!
2. Is ParticipantID (I'm assuming for the caller?) exposed as a built-in variable, or will it require another Data action to get it?
3. I see you mention using a second Data Action to stop the recording. Is there any way to detect when someone stops speaking, or perhaps timeout if they don't say anything?
4. I know it wasn't part of the initial ask, but I can see others wanting to play the recording back to the caller for confirmation. Can we do that too? Another DA, or using a Play Audio?

I appreciate your assistance!

------------------------------
Paul Simpson
Views expressed are my own and do not necessarily reflect those of my employer.
------------------------------


Original Message:
Sent: 06-11-2026 10:30
From: Jim Crespino
Subject: Recording Audio from a Flow

Hi Paul,

For #1, I believe it might be possible to develop a custom solution that would handle the use case.

To start with you need to record the caller's name.  You can do that today from Architect by making a Genesys Cloud Data Action request to call this API to start/stop the recording of a snippet of audio: https://developer.genesys.cloud/devapps/api-explorer#post-api-v2-conversations-calls--conversationId--participants--participantId--snippet-record.  That recording snippet will be a separate audio file saved alongside the entire recording in Genesys Cloud.

To play that audio file back to the agent at the start of handling the callback you would need to develop a small web app that would authenticate with Genesys Cloud using a PKCE grant OAuth client, use the Recording APIs (https://developer.genesys.cloud/devapps/api-explorer#get-api-v2-conversations--conversationId--recordings and https://developer.genesys.cloud/devapps/api-explorer#get-api-v2-conversations--conversationId--recordings--recordingId-) to retrieve the specific snippet recording, retrieve the audio file using the signed S3 URL provided by the Recording APIs, and finally use native Javascript to play the audio to the agent.  It sounds complicated but it is really straightforward to develop.  This web app can be embedded in an agent script using a Web Control and will run automatically when the agent script displays, or could be controlled via a custom button action.

If you would need help with any of this, Genesys Professional Services could consult & assist to help you with the development or they can build it for you.  If that is of interest to you then please reach out to your Genesys account manager.

------------------------------
Jim Crespino
Senior Director, Developer Evangelism
Genesys
https://developer.genesys.com
------------------------------