Presented by Bold 360 by LogMeIn

You have big expectations for AI-powered customer service tech, but is it the tech’s fault if they aren’t met? Catch up on this VB Live webinar to learn about the five biggest mistakes companies make when they implement AI and how to leap over these pitfalls and into real results.

Access the webinar on demand for free right here.

There are several common misconceptions businesses have around artificial intelligence, and these actually are the biggest barriers many have when it comes to implementing the technology, despite the clear evidence that it offers real, measurable results.

“At least initially, there was a general lack of understanding in the marketplace around what AI really is,” says Akhil Talwar, senior product lead at Bold360 by LogMeIn. “There’s a lot of this hype cycle phase that organizations have gone through. My hope is that we’ve now shifted into more practical applications of AI, but unfortunately that’s not always the case.”

Michael Butler, head of customer success at Ople agrees.

“What I’m excited about is the opportunity that AI presents to businesses across all industries,” says Butler. “But a lot of businesses don’t know how to get started with AI because they’re confused about what it truly is and how to dabble in it. My feedback is, just jump right in.”

Come to the table with some business questions you would love AI to address, he says, or at least try and solve for you, bring your data sets, and just start experimenting. Not all of them are going to work, and not all of them are going to be successful, Butler adds, “but you will find some spectacular insights once you start dabbling in it.” However, it’s this idea of “dabbling” that seems complex or inaccessible to the average business leader.

“If you’re a data scientist or you have a data science team, they understand the algorithms, they know how to get started, they know the process,” he says. “For me, as a leader in ecommerce and renewals [when I was at] VMware, I had a lot of data, a ton of data. I just wasn’t getting valuable business insights from it.”

It took sitting down as a team to discuss what answers, as business leaders, they wanted from their data sets.  One of the first areas was identifying the customers mostly likely to churn, which saved them billions of dollars.

Talwar notes that having a clear understanding of what AI can truly deliver is essential in this process.

“There are some risks associated with having misalignment within an organization, where they’re spending a disproportionate amount of resources on the wrong initiatives and they end up struggling to recognize an ROI,” he says. “One example is the confusion in the customer service and support domain around automation and AI being referenced synonymously. ”

You don’t need AI if you’re looking for simple automation. But on the flip side there are compelling use cases where organizations can create a richer customer experience, leveraging NLP and machine learning.

“You have to make sure you’re not paying AI prices for automation,” he says.

He points to chatbots as an example, which come in both these flavors. If you’re looking to collect contact information, or prospect and schedule a follow-up with some leads, you can set up simple automated process. Bots will ask a set of predefined questions, recall the answers, and block off calendars. But if you want to take and build a chatbot that behaves more like a human, where it can understand more broad sets of intents, be able to understand contextual differences, and handle a wide variety of topics in a single conversation, that’s where you really need AI.

That confusion around what AI really entails and what kind of technology you need often comes down to simply succumbing to the hype associated with AI.

“People try to fast-track anything that has an AI label slapped on it without going through the right set of motions,” Talwar says. “Very often, we’re seeing that there’s an executive internally that’s trying to elevate their position or trying to drive attention to some of their projects, so people will attach an AI initiative around it.”

It’s important to set the right expectations across the organization first, and get folks aligned on the specific business use cases.

“It’s great to jump in and start, but you want to start with the problem you’re solving, versus having this cool piece of technology or an algorithm you want to deploy,” he explains.

It’s also important that executives understand the cost and time investment involved, and to help address employee concerns around how an AI solution will impact their jobs – in other words, bring the right set of stakeholders together beyond just the people who are implementing it from a data science or engineering standpoint.

“Unless we can address the concerns of others, and be transparent with them, these initiatives come up against a lot of friction, and they’re less likely to succeed,” Talwar says.

Getting buy-in from stakeholders comes down to identifying which business pain points AI can help with, and looking at the return,” says Butler.

And then you get down to proving that out. The advent of AI as a SaaS platform really speeds up your experimentation life cycle and allows you to succeed or fail much quicker, so you can go back to your stakeholders with solid results and get buy-in to continue to push your AI success: say, identifying the nine percent of customers who are at risk for churn, addressing the issue, and being rewarded with higher renewal rates, and more cross-sell and upsell, for example.

It requires putting specific business metrics around experimentation, and people are failing with the belief that they need massive amounts of super-clean data right out of the gate, otherwise they won’t be able to get accurate results.

The reality is, you need a statistically efficient amount of data for the neural network to learn the predictions, Butler says.

“And it doesn’t need to be 100 percent clean, to be honest,” he says. “If you wait to clean it and you spend 80 percent of your time cleaning the data, you really don’t get to the fun part, which is making the models, tuning the models, and having the models kick out business predictions. Just start dabbling in it and playing in it, and you will learn. You’ll adopt it, because it’s the future.”

For the in-depth discussion around the kinds of use cases AI is creating tremendous ROI in, specific strategies for the specific hurdles you’ll most often encounter, and a look at the security issues around implementing AI, don’t miss this VB Live event!

Access on demand here!

You’ll learn about:

  • What AI actually is (hint: it’s not automation)
  • The importance of buy-in from executives and agents
  • How to approach AI implementation and measure success
  • The impact of AI across the customer journey


  • Akhil Talwar, Senior Product Lead, Bold360 by LogMeIn
  • Michael Butler, Head of Customer Success, Ople