Genesys Cloud - Developer Community!

Hi everyone,

We have a requirement in our operation to anonymize the agent's name in chat transcriptions and reports provided to customers.

Currently, the PDF export displays the full name of the representative. To comply with data privacy regulations (such as GDPR/LGPD) and to protect team's identity, we need to replace the agent's name with a generic identifier (e.g., "Agent" or "Representative").

I would like to ask the community:

1. Has anyone encountered a similar requirement and implemented chat transcription anonymization?

2. How did you structure this solution within your Genesys Cloud architecture?

3. Could you share any tips, best practices, or recommended approaches for handling this data transformation via the API?

Thank you in advance for your support and insights!

Hi Maria,

As far as I know, there isn't currently a native Genesys Cloud feature that allows us to anonymize or replace the agent's name specifically in the exported chat transcription/PDF.

I'm also curious to know if anyone in the community has faced a similar requirement and found another approach within Genesys Cloud.

One option I was considering would be to handle this outside of Genesys Cloud through a custom application. The application could retrieve/export the interaction data from Genesys Cloud, process the transcription before making it available to the customer, and replace agent-identifying information with a generic value such as "Agent" or "Representative".

This way, instead of providing the customer with the native Genesys export, we would provide only the processed/anonymized version generated by the custom application.

One important consideration is to anonymize based on the participant/user identity, rather than simply performing a text replacement of the agent's name. A simple string replacement could accidentally modify the conversation content if the agent's name happens to appear naturally in a message.

Of course, this introduces an additional layer to maintain, so I would be very interested to hear if anyone has implemented something similar or knows of a more native or simpler approach.

Hi @Maria Carvalho,

This is a very interesting topic. Thanks for sharing it with the community!

I did some research, and it seems that this functionality is not currently available natively in Genesys Cloud. From what I understand, if the goal is to export the PDF directly from the platform with the agent's name anonymized, this would require a native option in Genesys Cloud.

However, I believe the API below could be useful for building a custom solution. The returned JSON identifies the participants through the participantPurpose property, where internal represents the agent and external represents the customer.

You could process this data programmatically and replace the agent's name with a generic identifier before presenting or exporting the transcript to the customer.

https://developer.genesys.cloud/devapps/api-explorer#get-api-v2-speechandtextanalytics-conversations--conversationId--communications--communicationId--transcripturl

Thanks and regards. 

Hi Maria,

What you could do is use a portion of the blueprint below. Using Jhonotan's comment as well this enables you to process the data without an external system to create it. You could use a trigger or Execute Flow Data Action dropping in the Conversation ID. You could even have this in a script with a conversation ID and email box to send to a customer.

https://developer.genesys.cloud/blueprints/send-message-transcript-by-email/

Hi Maria,

One option is to use Genesys Cloud's native Sensitive Data Redaction feature. With personal information redaction enabled, the platform attempts to automatically identify and mask names and other personal data in transcripts. It's worth paying attention to permissions, as users with Recording > ViewSensitiveData can still access the original, unmasked data.

Since this identification works on a "best effort" basis, there's no guarantee that every name will be detected. Another alternative would be to use a generic display name for agents, such as "Agent", although this may not be very practical depending on the operation.

For scenarios that require greater control, I would also consider processing the transcript through the API before making it available to the customer.