AI Script Recommendation

[jovieticar101]

Would love to have Auto script recommendation for the agents. Example: I have a call, the system will capture the incoming voice then calculate it then provide a good script / rebuttals for our agent base on what we've capture.

My Question is how can we capture the realtime voice of a person from the other line.

[mflorell]

We have helped a couple clients set this up, it isn't simple and it certainly isn't cheap, and even after all of the technical parts are set up, the training can take months.

[carpenox]

i too have helped a few clients with this, i can second what matt said, it isnt an easy process, definitely it not cheap and does come with incorrect info still as AI just isnt quite there yet.

[williamconley]

... the system will capture the incoming voice then calculate it then provide a good script / rebuttals for our agent base on what we've capture ...

My Question is how can we capture the realtime voice of a person from the other line.

That sounds like a best-use-case scenario for the present state of "AI" (note the quotes, lol).

You want the AI to provide "best guess" rebuttals for the agent. That's very good, actually.

Yes, You can have an AI "monitor" line which could then provide information via a scripted iFrame which renews/refreshes every second to get "up to the second" input from your AI, assuming you can handle the programming of the AI.

Two stages: Setting up the monitor line at the start of the call so the AI can process the audio and generate prompts/options for the agent. Then coding the iFrame into the script with AJAX in the iFrame to continuously refresh from the AI service's HTML production (or text response, but then the Vicidial or other web server would have to accept the AI output, convert it to something visually functional for the agent, and then reply with that information when the next refresh occurs).

Absolutely possible with today's technology. IF you can write the AI code to interpret and provide the rebuttals. Presently, however, we just use a soundboard with all the rebuttals in the agent's voice premade and let the agent push the button corresponding to the audio (recorded previously, with proper intonation and practiced terminology) into the agent session. Vicidial's soundboard is very useful for that.

[carpenox]

i have used openai for ai amd as well as ai bots for clients and a service we offer, if youre interested in a free trial hit me up

[jovieticar101]

... the system will capture the incoming voice then calculate it then provide a good script / rebuttals for our agent base on what we've capture ...

My Question is how can we capture the realtime voice of a person from the other line.

That sounds like a best-use-case scenario for the present state of "AI" (note the quotes, lol).

You want the AI to provide "best guess" rebuttals for the agent. That's very good, actually.

Yes, You can have an AI "monitor" line which could then provide information via a scripted iFrame which renews/refreshes every second to get "up to the second" input from your AI, assuming you can handle the programming of the AI.

Two stages: Setting up the monitor line at the start of the call so the AI can process the audio and generate prompts/options for the agent. Then coding the iFrame into the script with AJAX in the iFrame to continuously refresh from the AI service's HTML production (or text response, but then the Vicidial or other web server would have to accept the AI output, convert it to something visually functional for the agent, and then reply with that information when the next refresh occurs).

Absolutely possible with today's technology. IF you can write the AI code to interpret and provide the rebuttals. Presently, however, we just use a soundboard with all the rebuttals in the agent's voice premade and let the agent push the button corresponding to the audio (recorded previously, with proper intonation and practiced terminology) into the agent session. Vicidial's soundboard is very useful for that.

I dont need the sound board. my plan is to give only the best script/rebuttal for the agents. all they need to do is to read the script until they close or hangup the call.

the reason is to expand our talent pool cause all they need to do is to read the script

[williamconley]

I dont need the sound board. my plan is to give only the best script/rebuttal for the agents. all they need to do is to read the script until they close or hangup the call. the reason is to expand our talent pool cause all they need to do is to read the script

Seems quite reasonable. I was really suggesting the SoundBoard more for the fact that it's a custom-entry-in-an-iframe already. If you used the AI to generate something similar ... you could have your solution almost premade.