Genesys Cloud - Developer Community!

 View Only

Sign Up

  • 1.  Building an experimental AI-based CX journey

    Posted 10-07-2025 18:08
    Edited by Lucas Woodward 10-07-2025 18:59

    Last night I sketched out an experimental CX journey I'd like to create, in a LinkedIn post.

    It is:

    1. Start a conversation with a text-based chatbot about selling my books on my mobile phone (aka cell phone)
    2. Have it transfer to an AI powered voice + video bot to discuss which books on my bookshelf to sell, and potential cost of each
    3. Once decided to have this information fed back into my existing chatbot session, to allow me to continue

    The sketch I shared:

    My main reason for doing this is to experiment with whether this is technically feasible using Web Messenger, and Google's Gemini Live API - two great technologies I'd love to see working together.

    It'd be great to get thoughts from the community on how they'd approach this. I am thinking:

    • Start with the official Web Messenger experience - rather than the Web Messenger Guest SDK
      • This is for simplicities sake, although may require unacceptable workarounds
    • Use the Web Messenger's Database Plugin to allow the Inbound Message flow/chatbot to 'notify' the website when to start Gemini Live API I'll just use a Data Action for this actually.
    • The Live API will have 2 tools
      • Search for ISBN by book title - it can use this when being shown a book
      • Agree sale of book - this will cause the book's ISBN to be sent to the Web Messenger chatbot via the sendMessage command
    • When customer/Live API finished the conversation in Web Messenger has the books (see sub-point above) and I can send a message to the conversation to prompt it to continue the Web Messenger based conversation.

    I'm sure I will make many amendments (and mistakes) along the way, but does anyone see any glaring issues with my thinking?


    #MobileMessenger
    #PlatformSDK

    ------------------------------
    Lucas Woodward
    Winner of Orchestrator of the Year, Developer (2025)

    LinkedIn - https://www.linkedin.com/in/lucas-woodward-the-dev
    Newsletter - https://makingchatbots.com
    ------------------------------



  • 2.  RE: Building an experimental AI-based CX journey

    Posted 10-14-2025 16:30

    I thought I'd share an update. I have finished the first iteration of the video/audio AI part of this solution. I have created the ability for the 'customer' to interact with the live audio/video AI so it can recognise a book via its ISBN in barcode:

    Video and explanation in my LinkedIn post: https://www.linkedin.com/feed/update/urn:li:activity:7383263424652259328/

    In more detail, it works by:

    • Using the bowser's native ability to scan barcodes against the live webcam stream. This is better quality and with a quicker frame-rate than what my server receives to send to the AI
    • Barcodes detected above are immediately sent to my server
    • ISBN in barcode used to pre-emptively retrieve book details server-side, reducing latency in conversation
    • Gemini Live API given a tool to 'read_barcode', making conversational agent sound like it is indeed reading the barcode. But tool returns result of pre-emptive search

    What's next:

    Next I want to focus on having a Digital Bot in Genesys Cloud trigger the audio/video appearing for the customer, and receive scanned books it can use in the chatbot's conversation.



    ------------------------------
    Lucas Woodward
    Winner of Orchestrator of the Year, Developer (2025)

    LinkedIn - https://www.linkedin.com/in/lucas-woodward-the-dev
    Newsletter - https://makingchatbots.com
    ------------------------------

    Original Message


Related Content