Freshworks optimizes customer service interactions through genAI, sentiment analysis

00:00  
Hi everybody. Welcome to DEMO, the show where companies come in and they show us their latest products and platforms. Today, I'm joined by Payal Patel, she is the head of solution engineering at Freshworks. Welcome to the show, Payal.
 
00:10
Hi, it's nice to be here.
 
00:11
So what are you going to show us today?
 
00:13
Today I'm going to show you our Freshworks customer service suite. It's a modern AI-powered solution for customers and employees.
 
00:19
Alright, there's always a requirement that we have AI in here. So who is this designed for? I've imagined since it's customer service. You're talking about companies that have products and they have customer service platforms.
 
00:29
Of course, it's for companies of all sizes, from Main Street to Wall Street, all different types of industries, whether it's from healthcare all the way up until retail.
 
00:38
What problem are you solving that other companies haven't done, or that maybe companies haven't seen until now?
 
00:44
Three key things, one of them is all around fragmented channels of communication, lots of different systems, inconsistent customer experiences that [the systems are] causing. Second, it's all about automation. How do we ensure that agents can be more efficient? And then finally, supervisors and leaders, they lack insights into the business, and we're providing those insights to those leaders.
 
01:06
Some very cool stuff. So what would companies do if they didn't have this? I would imagine there would be an awful company because they didn't have this and fragmented. I like that word because it means, like, whenever I'm on the phone with someone, and then I get transferred to another person, I have to re-say all of my information. And you're eliminating a lot of that, right?
 
01:21
Yeah, we are eliminating a lot of that. And if they continued with it, they'll continue to have more disparate systems where agents really kind of get a holistic view of their customers and know what's happening, also, agents get inundated with lots of mundane tasks that they could maybe get optimized through AI. So those are some of the things cool.
 
01:40
Let's go right into the demo and show us what you got.
 
01:42
Sure, let's get into it. Okay, so I'm going to be walking you through a fictitious company today called Real Lux. They're a premier consignment store that provides luxury designer clothing and accessories. Our customer, Elise, is a marketing executive from Las Vegas, and she's recently purchased a luxury designer watch. Real Lux is really proactive in nature. They've already reached out via WhatsApp to check in on her to ensure that everything was set up to her expectations.
 
02:12
But for the case of this, something went wrong. Something always goes wrong.
 
02:16
Unfortunately, when she opened that Cartier watch, she sees scratches on the surface of the watch and decides to reach out. She contacts Real Lux on WhatsApp, which is powered by Freddy, our AI assistant self-service tool. Freddy immediately identifies her intent and seamlessly surfaces up Elise's recent order visually showing an image of the Ballon Bleu watch that she ordered. Elise confirms that that's the item that is giving her problems right now. And Freddy then goes ahead and prompts her to describe the issue via a photo or a video. Now, typically, customer Elise would normally have to explain the situation with a long sentence. Instead, here she can just easily go ahead and send off a video. This is where the magic happens. Freddy analyzes this video, and based on the analysis, is going to recognize that the watch has scratches on the surface. Now before Freddy goes ahead and says she needs a replacement, he's going to ask her to try to help Elise solve this issue by cleaning or using a microfiber to clean the surface of the watch. Now customer, Elise takes that advice. She really tries, but unfortunately, it didn't work. She's going to let Freddy know that that didn't help. Freddy's continuously going to help Elise, and informs her two options that she has. At this point, she can either replace the watch, or she can get a new watch arranged for repair. Elise thinks that the replacement would be a better option. She goes ahead and lets Freddy know that.
 
04:04
So Freddy is also analyzing the text messages that that she's giving to him too, right?
 
04:09
Yeah, it’s understanding all of that through natural language processing. Okay? Now, Elise is a little concerned. We have another problem. It's not going to arrive until June 21, and she needs it a lot sooner, right? So she's frustrated. I don't know about you, but when I'm frustrated, I start to speak in my native language a little bit and spewing out words.
 
04:27
So she's going to go ahead and reach out, letting Freddy know that this does not work for her, and she wants to be able to under get something done sooner. Now, Freddy's really smart. He's understanding that Elise is frustrated. He understands that there were multiple sentences in different languages as well, and responds in multiple languages as well, and provides her an alternative by transferring her over to an agent.
 
04:51
Do a lot of do a lot of chatbots now have that capability? Oor is that something that's really new?
 
04:58
Yes, that’s something that's newer. I think generative AI has helped with that, but we're actually making it even better with being able to understand the content.
 
05:04
So now we're going to go to a human right?
 
05:05
Yep, we're going to go over to agent Ethan. So let's talk about what it looks like for agent Ethan. Agent Ethan works out of his unified agent interface, where he can see all of the conversations and tickets that have been routed to him. This could be under WhatsApp. This could be from SMS, Instagram, Line, many other channels of communication now, Agent Ethan really wants to prioritize the least happy customer first
 
05:30
Angry customers, you gotta get them done first. How does does the system understand the the sentiment? Like he, how can he tell like, how does the system know that Elise is the most upset?
 
05:39
It's reading the context of the conversation back and forth through the bot, and it's looking for keywords that would detect some form of frustration.
 
05:47
So when she said I'm frustrated or use the angry emote, right?
 
05:51
So I'm gonna go ahead and go ahead and sort that by the least happy customers, and we'll see that Elise, our customer, is right at the top now as he goes ahead and reviews the content at the top, we see that Freddy copilot, our AI assistant, helps to summarize the entire conversation, and this allows agent Ethan to get up to speed as quickly as possible. He goes ahead and responds initially with to Elise, and as he tries to send this message, our proactive quality coach catches all of his spelling errors. Generative AI is understanding that there are typos, and Freddy co-pilot comes to the rescue, making agents more productive, ensuring to correct those typos that agent Ethan would have typically made, right?
 
06:41
Because if Ethan had sent that out, it could be even more frustrating for Elise, right?
 
06:46
Exactly, it could actually hurt the brand, and that's why we want to get ahead of these. So we'll go ahead and fix some of those typos and then send the message out to Elise. Next, Agent Ethan is going to review all of the information around Elise. He wants to get up to speed on what she who she is, and what he can do for her, so he reviews the customer information. Can see some integrations through Shopify, even his order details through third party systems like Stripe. Now Elise, on the other side, is getting frustrated as always, and she comes back and gets impatient and says, can you just share and expedite the shipment of this? Watch over to me. Agent Ethan knows that he has a knowledge article about expedited shipping, and he's able to go ahead and share that there are some expedited delivery options. Goes ahead and sends that off to Elise. And finally, at least Elise acknowledges that that works for her. She says, Yes, great. We finally really resolved all of Elise's problems. Ethan can quickly go ahead and resolve that conversation and move on to the next customer that he needs to help for the day.
 
08:01
All right. We have one more persona I want to show. Elise will get her watch right?
 
08:06
She's going to get her watch in time for the cover.
 
08:08
You have one more thing that you want to show. This is really cool, too.
 
08:10
Next, let's shift our attention to customer service leader Adam. Customer Service leaders like Adam used to spend considerable time shifting through reports to grasp the top customer issues, and here, Freddy Insights is able to surface up those insights. Within a matter of seconds, he can ask Freddy show me high frequency contact scenarios, and immediately, Freddy can analyze the data and share all of the top inquiries that are happening within the customer contact customer support organization.
 
08:42
So you're getting  a lot of requests for returns and replacements here.
 
08:46
Now before Gen AI, do you know what Adam used to have to do? What did he do? He would have to create attributions for all of the fields. He'd have to understand, have some BI guy help him figure out how to create the report. And then finally, he'd have to maybe export the data and really make that analysis, right? And now this is becomes way, way more simpler. So as he's looking at this, he knows he needs to fix the returns and replacement problem. He's going to find out what queries, what are some of the sample inquiries that customers have been asking about this? Okay? He's able to see that there are two or three different things going on that could be self-serviced, and he knows that he needs to fix this so that he can take some pressure off of his team. He's going to say, let's go ahead and create a bot flow for these scenarios. Freddy asks a simple question, where would you like this deployed on? Which channel? Obviously we can deploy it on all. But for now, we're just going to deploy it on WhatsApp. And then finally, Freddy copilot is able to create that bot flow for atom, reducing that administrative effort. And guess what? Configuration has become easier than ever.
 
09:53
That came up pretty quick too. Yeah, it did. Yeah, all right, so there's a lot of features here that, and I'm sure you've got a ton of other features. So. Where can people go for more information, to learn more about this platform?
 
10:04
Come to Freshworks.com and learn about all our products from there.
 
10:07
Payal Patel, thanks to Thanks for joining us and thanks for the demo. That’s all the time we have for today's episode. Be sure to like the video, subscribe to the channel, add any thoughts you have below. Join us every week for new episodes of DEMO. I'm Keith Shaw, thanks for watching.