Freshworks took over Arsenal’s Emirate’s Stadium in London last week for the Refresh Europe event, with partners, customers, prospects and CXM in attendance.
The main focus of the event, aside from the value Freshworks provides through the ease-of-use in Freshdesk and Freshservice to customers, was the arrival of its second-generation AI. Freddy: Agentic AI is badged as the “Uncomplicated AI that delivers…” With CEO Dennis Woodside taking the stage to highlight the company’s success and upcoming products.
Say Hello to the New Agentic Freddy AI
The Freddy AI Agent, has been in testing since October but is now available across the Freshworks user-base, providing a more active agent who can learn and resolve problems through continuous learning.
The no-code element to Freddy AI Agent Studio means that citizen developers can create and deploy customer agents to solve specific issues without waiting for developers to become available.
Mika Yamamoto Talks Up Agentic Freddy
The new Agentic AI Freddy can also perform financial tasks, such as providing refunds or loyalty rewards, and updating business records. All helping to reduce the workload of repetitive minor tasks for customer service agents, while cutting resolution times and boosting overall service quality.
Freshworks Chief Customer and Marketing Officer Mika Yamamoto took time at the Refresh event to sit down with CXM to discuss the road to Freddy’s Agentic launch, and how customers are adopting and adapting to him.
How easy do you find it to market Freshworks to a business audience that demands simplicity, leading to the “Uncomplicated” campaign?
The reason why we ended up going down the road of creating the Uncomplicated campaign is that we saw a trend with our customers telling us why they actually buy Freshworks and why they stay with us. They switched to Freshworks from the CX standpoint because they feel like we’re less complicated, and we can deliver; we say what we mean, and we mean what we say. And then what we deliver is in fact what our customers get, and that they’re able to implement it faster for less money, and that our pricing models are actually a lot more predictable in terms of what they pay.
For all those reasons, customers tell us why they do business with us. And so we’ve essentially taken, from a marketing point of view, what our customers tell us, and then we repeat it for our potential customers. We address these pain points, using terms that real people use rather than vendor jargon. And so while we want to be less complicated in our software, we want to be less complicated in the language we use. And we found that it’s actually resonated quite well.
You use Freshworks internally, are there any moments where it really impressed you?
As Chief Marketing Officer, I also run support. And so we use our products to provide support. The wow moment for me was that I can, as a person that’s not usually a customer support person, go into our ticket queue, pick a random ticket and click on it, and Freddy AI Copilot will give me a recommended response based on our knowledge database.
It’ll give me a recommended response for that customer. It’ll personalise that response. It’s already read the ticket and come up with a draft response so that when I open the ticket and look at the response, I can very quickly say, this matches or it doesn’t.
I can tweak the language a little bit to sound more like me. And then I can send it and I can do that within the span of a few minutes for most tickets, compared to a previous area of around 15 to 20 minutes.
I’m sure it takes our customer support people less time. But what struck me was how empowered I felt. If you’re a customer support agent able to respond quickly to a customer accurately, and increase customer satisfaction, that is powerful. So the wow moment is how effective the Freddy AI Copilot is at helping people with their jobs.
And in terms of new Freddy Agentic AI… it has been in testing for a while, so whats some of the more exciting feedback you’ve heard?
The exciting feedback that we hear is twofold. Number one is how fast it is to get up and running. So we’ve had customers that within hours, what they can do with our conversational AI solutions is point our AI agents to a data source, be it a PDF or website or what have you. And it ingests that information and can immediately start responding to customer queries.
The new release that we announced today is that with the Agentic features, it can actually take action so it can look up an order’s status. It can cancel, update or perform other tasks with the order. And so the wow moment is the number of tasks it can handle and help reduce mind numbing tasks. I mean, no human wants to actually go cancel orders, update orders, search on orders.
They want to talk to customers and help them with complex problems. And so the wow moment is how fast it is to get up to speed making working as a customer support person more satisfying because they don’t have to do the mind numbing tasks, and they can actually use their brains and use their use their human skills and abilities to connect with customers and help them with complex issues, which is more satisfying for them.
In terms of business concern about AI and reputational, consequences, Freddy delivers responsible and high-quality AI. How confident are you in that?
I think that from a reputation standpoint, there are various areas where that can get damaged. The worst case scenario is where there is a data breach, where they’re sharing information they shouldn’t share about their customers. And that’s where I feel very confident that from a Freshworks point of view, we take very stringent measures to make sure that our customers’ data, and employee data are very well protected.
We’ve been in the AI business for over six years, we’re not new to AI, we’re not new to security. This means we feel very confident in terms of how we’re protecting employee and customer data, which we do with clear guidance for customers and users.
The next area where you have a potential for reputational risk is in actually providing the wrong information or bad data. And what we’ve found is if you (a customer has) bad data to serve your data, your AI engine, you’re not necessarily going to get great results back.
Insert bad AI joke here…
The next reputational area is in tone. So, we tested an AI model that sent customer mails automatically. Before letting it go into the wild, we were looking at the emails it was writing, and it was trying to make jokes. And they were really bad jokes. They weren’t off-colour or anything, but they were actually just not very good jokes. And so the tone wasn’t great.
If I were to receive these emails, I’d probably delete them, which, from a reputational standpoint, if something isn’t in the right tone, it can be off-putting and it can actually hurt your business because people won’t respond to your email. So I would say that the third category is making sure that you can have the right tone and along those same lines, if you’re dealing in multiple languages, having accurate translations.
So those are the areas where I feel very confident about where we sit on security. We have to make sure that we feel confident about how we’re educating our customers in terms of how they feed their models to be able to be contextually relevant for their customers and continuing to look at how we actually manage tone. We allow for, and we enable our AI agents to be able to say, do you want a professional tone? Do you want conversational tones? And we’re continually looking at how we perfect that.
Where do you stand on the AI vs human jobs and roles issue?
That’s a common question I get from financial analysts and tech analysts. I don’t see a lot of my peers saying, we’re going to wipe out all human agents. We want to actually move humans to deal with what only humans can deal with. We’re not looking to wipe out our entire support force. The value of their communication is only going to increase, which makes the service better, and if you are a customer talking to a bank or a cable company, you sometimes need to talk to a real person.
What we do see is that operating expenses as a result of servicing customers is actually growing less fast than the growth of a company. That’s another reason why it’s not happening.
There was a lot of presupposition two or three years ago, that when AI started to get more popular people would no longer use the phone. Interestingly, older generations want to use the phone. Younger generations also want to use the phone. And so telephony isn’t going away, and a huge point of feedback that we got from all of our customers in all countries over the last several weeks is that telephony is something that they want us to continue to invest in, because people do want to pick up the phone.
We’ll have a number of conversations with Freshworks customers over the coming days to see how they are dealing with AI, CX and other issues, check back!