AI Call Summaries Generic vs Custom LLM | AI Call Center Software

Your company has a call center, and you’re wondering if AI call summaries can help eliminate wrap time. I’m here to tell you they can help only if they’re good, if they’re not good agents have to proofread them which equals more time.

In this video, I explain what needs to happen to make it good vs bad.

Want my recommendations on the best call center software vendors? Click the button below and ask me today.

About Me

Mike Smith has been helping companies select the best telecom, WAN, security, and cloud services since 1999. He founded AeroCom in 2003, and has been the recipient of numerous business telecommunications industry awards, including being recognized as one of the top 40 business people in tech-heavy Orange County, CA. Follow Mike on YouTube, LinkedIn, Reddit and SpiceWorks.

Transcript

When it comes to cloud call center software, your company might have heard of something called A.I. Call summaries. What that is, is after every call, a A.I. gives a summary, a written summary of that call, including most important things covered. The takeaways, the action items after the call. All that good stuff, right? Objections or, you know, things that they brought up that were problems, questions that were asked, products that were discussed, that type of thing. So sounds great, right? That is, if the A.I. call summary is accurate and good. If it is not, then what happens is agents either have to review the summary after the call and edit it. So what’s the point if they’re having to go in and edit it and take time, take wrap time to edit the call summary, or you know, if they’re not good and nobody edits them, then future

folks talking to customers aren’t even going to look at that information the next time they talk to a customer. So the customer calls back, you know, a month or two from now, they go to look at the call summary and they just don’t pay attention to it. They’re going to have to ask the customer all the same questions all over again because they know that they can’t rely on that. So obviously, you want to make sure your company is getting good AI call summaries, not bad call summaries.

Well, what’s the difference?

The difference is the large language model, the LLM, that the cloud call center software is using to power. The A.I. calls them summaries. Are they using a generic large language model? like GPT three or are they using a custom large language model? So that’s the main difference. And I’m going to explain about that in this video. So we’ll get into that in a second.

My recommendation

But first, if you’d like my recommendation on the best cloud call center software vendors, your company should be quoting reach out, send me an email or give me a call (714.593.0011). This is what I do. I’m a broker for all the major cloud call center vendors out there, and I will help your company make a lot better decision in a fraction of the time.

Don’t start googling it and try to search all the different vendors out there. There’s way too many. You probably end up with the wrong one. Instead, just reach out and contact me. Trust me, it’s the best way to go. More information on that at the end of the video. Also, don’t forget to hit the like button down below. Ring the bell and subscribe to the channel so you don’t miss any of my feature videos. I put out a new video at least once a week, if not more, and that’s the way you’ll make sure you don’t miss anything and subscribe to the channel and leave a comment down below. Tell me what questions you have on this video and tell me what future videos you’d like me to do.

Generic vs Custom LLM

Okay, so as I was talking about in the beginning of the video, why is it important that a cloud contact center software vendor uses a custom large language model as opposed to a generic large language model like GPT?

1. Errors in transcription

Well, the first thing you want to think about is, call center conversations are just going to be a lot different than the normal written word. Like if somebody is typing something out, there’s a it’s that’s a lot different process than a conversation. For instance, translating what people are saying, be able being able to understand the different ways people speak as opposed to the different ways people write in. Translating that from speech into writing is obviously more challenging than just trying to trying to use a large language model off to somebody typing in a question.

2. Disfluencies in speech

Also disfluencies in speech like you might be hearing this video when I’m talking like using the word or or maybe talking too fast, things like that. That’s something that only happens on a phone call.

3. Background noise

Another thing is background noise. So if either the person calling in, maybe the customer or the agent is in a room where there’s background noise, is the large language model able to overcome that?

4. Agent vs Customer dialog

Lastly, agent versus customer dialog? Is the large language model able to really determine who’s talking and why that’s important? Is the agent talking or is the customer talking, and how is that translating into the call summary? So all those things are really important for specifically for phone calls taking place in a call center as opposed to just a generic large language model like chat GTI where people are typing stuff in on their computer using a browser.

Remedy: Machine learning/training

So how do vendors create their own will? Basically the cloud contact center vendors who create their own use between the multiple agents, They’ve come up with over 2 billion conversations they’ve used to help develop their custom large language models. So they listen to tons and tons and tons of conversations and use machine learning and people to help fine tune their large language model specifically for contact center calls as opposed to other forms of generic large language models.

So that’s how they’ve done it. They’ve listened to billions of conversations and they’ve edited the way their large language model works and customized it custom tailored it just for context. Center conversations.

Zero wrap time

And at the end of the day, the result of that is a large language model that has less errors and creates very good ai call summaries. So that’s why it’s important If you’re looking for cloud contact, center software with call summaries, you want to dig a little deeper and find out are they using a custom large language model or are they using just a generic large language model like Chad? GP And just trying to say, oh, it doesn’t matter. They’re all the same because they’re not all the same.

Still confused? Reach out and contact me

I hope this was helpful. If so, reach out, send me an email or give me a call (714.593.0011).  if you’d like My help on which cloud contact center software vendors your company should be quoting for things like air. I’ll help you quickly determine which vendors your company should be quoting or introduce you to the best sales reps at those companies. And then I’ll oversee the coding process to make sure your company gets the best pricing from your vendor of choice. I’m a broker for all the major cloud contact center software vendors out there. I’ve been doing this for over 20 years, so I know the industry inside and out and there are literally hundreds, if not thousands of options for you, way too many to make a good decision on your own. So reach out and contact me. I’ll help you kind of sift through all the options. And based on your companies requirements, I’ll make sure you’re quoting the best vendor with the best reputation so that your company can make the best decision in a fraction of the normal amount of time. And the nice thing is, is the cloud contact center software vendors pay me my broker fee so your company doesn’t have to even pay me to ask me for my advice. So there’s no reason to at least not reach out and see what I have to say.