How to make ZVA's AI call center by Voice agent

Hi Hiro!

We could capture steps 3-6 as part of a ZVA skill & potentially use a tool to then email this on. All very possible.

If you build with ZVA Voice Agent 2.0 (or 3.0) then you wouldn’t need intent recognition, just well written prompts & guidance.

​@Hirobumi Chamoto ​@Chris Holt 

Make a skill that is separate from the skills collecting #3-6 and then create a “ticket by email” integration support channel.

The new skill needs to be “before ending call, submit a ticket.” In the trigger descroption, you will write “before ending the engagement you MUST submit a ticket with the summary of the conversation using this skill.” 

The instructions in the new skill would be: 

Once the customer has finished the conversation, submit a summary of the conversation using the tool / (create_ticket). Once you have submitted the ticket you can /tool (end_conversation). 

In the tool “create_ticket” that is a built in tool, you will get a new screen to custom the action. Select the “ticket by email” option under Integrations and select “include transcript with ticket.” See image. 

This will submit the transcript of the conversation (#3-6) to the email address. 

If you do not want a transcript, but instead want a summary of #3-6, then tag me back in the this thread. 

___________________________

Question 2: AIC 2.0 or 3.0? 
There is a field in the AI Studio, Virtual Agent configuration that says “AIC 3.0” and it has a radio button for “on/off.” If the toggle is off, then you are using AIC 2.0 orchestration. If you the toggle is on, then you are using AIC 3.0 according to the volume slider that appears. AIC 2.0 or 3.0 is the orchestration tooling. For the specific LLM processing through the orchestration layers, that is the “model” field below the AIC 3.0 field. You can select a model from there. If you do not have model selection show, as in the picture, then you do not have this permission due to your license type. Contact your sales rep to get the full version of ZVA+. 

Hi Hiro! 👋

Thanks for sharing your detailed scenario. From what you described, it sounds like you’re looking to set up a ZVA (Voice Agent) workflow for your call center and also track conversations via email. Here’s a structured way to approach it:

1. Call routing: Configure the IVR to allow language selection (e.g., 1 for Japanese, 2 for English).
2. Voice agent conversation: ZVA listens to the customer’s situation and asks follow-up questions to gather details.
3. Solution suggestion: ZVA provides automated solutions based on your knowledge base.
4. Escalation: If the solution doesn’t work, the agent collects customer info (name, mobile).
5. Email summary: Check if your ZVA platform supports conversation logging via email often this requires enabling conversation transcripts in the settings.
6. Knowledge base & intent management: Regularly update your KB and intents to improve accuracy. This can include adding phrases, example queries, and responses.

Since you’re new to ZVA, I recommend going step by step, testing each stage with sample calls before full deployment. Also, check the platform documentation or community tutorials they often include screenshots for setting up IVR, logging, and intents.

Hope this helps! 🙌

​@Chris Holt -san, sorry late my response. I watched your Youtube.

Flow using ZVA 3.0(our version).

1.Made voice agent

2.Japanese voice set

3.Uploaded knowledge

4.Set skill
5.IVR for ZCC
6.TEST -> release
I know from step 4,5,6 tomorrow.

What shall I do next?

I really appreciate your dedicated supports, Hiro.

Your scenario sounds possible with ZVA, especially for a multilingual voice support flow using IVR + AI voice agent features.

From what I understand, you would likely need:

• IVR menu setup (Press 1 for Japanese / Press 2 for English)

• Speech recognition + intent detection

• Knowledge base integration

• Conversation logging/history

• Email notification or CRM integration

• Human escalation flow if AI cannot resolve the issue

For steps 3–6, the Voice Agent should be able to collect customer intent, ask follow-up questions, and transfer structured conversation data. Step 7 may depend on whether your current environment supports conversation transcript export or email integration.

Since your Widget menu is blank, there may also be a permission, license, or setup issue affecting configuration visibility.

As a beginner, I would recommend starting with:

1. Language/IVR setup

2. Intent creation

3. Knowledge base linking

4. Basic conversation flow testing

5. Escalation + email workflow

Hopefully someone from the community or ZVA team can share a more detailed implementation guide for your use case. Good luck with your Japan support project 

​@James Smite san, I appreciate your summary and cleared point.

 Exactly.

I’d like to make the menu step by step immediately.

But I have not solution.(I tried to ChatGPT and  Copilot but I can’t get any detail action list).

1.Language/IVR setup
2.Intent creation
3.Knowledge base linking
4.Basic conversation flow testing
5.Escalation + email workflow
 

I will look for them continually.

Thank you.