MarTech Interview with Fergal Reid, Director of ML at Intercom

Fergal Reid, Director of ML at Intercom chats about the future of AI and what marketers need to do prepare for an AI driven marketing ecosystem:

_______

Welcome to this MarTech Series chat, Fergal – tell us about yourself and more about your role at Intercom…

I’m the Director of Machine Learning at Intercom based in Dublin, Ireland. At Intercom we’re building an AI-powered, automation-first customer service platform. We help businesses deliver faster customer service, which keeps customers and teams happy and costs low. I’ve spent the last six years leading Intercom’s Machine Learning function.

We’ve recently built and shipped several new AI products within our platform, including three automation-first features built using OpenAI’s GPT-3.5 technology within our Intercom Inbox. We’ve also just built Fin, our breakthrough AI customer service bot built on GPT-4 technology.

When it comes to automation and the future of customer support and customer success: how do you feel both humans and automation should tie in together to create exceptional experiences without it being a case of one over the other (too much automation can after all affect the essence in a customer journey)?

At Intercom, we’ve long believed the future of support is automation-first, but not automation-only.

In today’s fast-paced business environment, AI should be the first responder. It can successfully resolve most commonly asked customer questions, and for many businesses, that is the majority of customer inquiries. With the latest AI models, it is feasible that we will see AI accurately and immediately resolve nearly half of customer questions without needing human involvement. But if AI cannot answer a question, it will seamlessly handover to a human agent, passing on all relevant information in the process. This is going to give customer service teams way more time to address complex, nuanced customer queries, while providing faster and more accurate automated responses to those who do not want to speak with a human.

We will see AI and automation take on the repetitive, rote work that human agents typically handle, but humans will always remain an important and necessary part of the customer support experience. In fact, their job will get even more interesting as they handle only the most complicated questions, high-value customers, or even move into newly formed support roles created by AI, like conversation designers.

There isn’t a simple human vs. AI narrative that you might assume. Customer support teams generally see high staff turnover and constantly have open roles. We think with AI the customer support role will probably become more interesting, and that managers will have to invest to upskill and grow their reports.

Marketing Technology News: MarTech Interview with Sanjay Mehta, Head of Industry, Ecommerce at Lucidworks

Can you talk about some leading global brands and how they drive their customer support process with the support of AI – what key takeaways can other businesses learn from these formats?

AI is still very new, and everybody is still trying to figure it out. But, here at Intercom we’ve seen some really cool startups use our products. Gifted’s Customer Success Manager, Barak Elisha, told us “The new AI Inbox features are awesome! I’m really excited about the summarize option, which makes it easy to summarize long conversations and saves us a ton of time.” Another positive reaction we had was from Robin Salimans, CTO at Luna: “The AI Inbox summarization feature is great for catching up on conversations my colleagues have had with users, and also for quickly entering a summary into our bug tracking system. Great stuff!” Overwhelmingly, we’ve been pleased with the way AI has helped support teams save time and serve their customers more effectively than they could before.

To businesses looking to implement AI, it’s most important to first consider whether AI is a fundamental change for your industry—really think from first principles about whether it could make your product quicker, easier, faster, or more productive. If so, you need to start investing now.

Once you identify a promising area, start small with a high-velocity project that has a clear goal and delivers real value to customers. At Intercom, we built our first AI feature to provide human support agents ways to condense, summarize and expand on the answers they were already writing in conversations with customers. We took an existing workflow and augmented it in a way that was almost strictly better. If it didn’t work, there was minimal downside to the end customer. If it did work, it made things better and faster.

Next, it’s important to ensure you have strong cross-functional collaboration between all of your internal teams. At Intercom, our ML team was in charge of identifying the opportunities and building initial prototypes. Once that was done, we had incredible executive stakeholder support to overcome internal obstacles. With alignment on the end goal across all the entire team, you can execute innovations quickly and seamlessly.

Can you talk about the key areas in which you are seeing AI not being used sufficiently enough to drive impact and scale in certain customer facing functions (marketing/sales) and what you’d tell teams to do more of to fix this lag?

Things are so fast moving at the moment, in that we’ve just seen a massive capability unlock. The technology has just jumped forward, so that honestly probably most areas are underutilizing it. There’s also so much going on that it’s hard to keep track of all the new offerings. I would encourage companies to do some of their internal experimentation if they have a software development function, and also keep a close eye on vendors in their space that are rapidly adopting AI, as well as new entrants. There’s huge potential in this space—it’s almost a full time job to keep track of all the new features, and figure out which ones are worth adopting. But it’s worth investing in doing this, because so much value will rapidly unlock, and it’s easy to miss it if you aren’t keeping pace.

What are some predictions you’d like to share for the future of AI and its impact on marketing, sales and customer facing instances like support and service?

With the advent of AI, the customer service job will get much more interesting and rewarding. Support agents will only handle more complex and interesting questions that require a human touch or VIP conversations. The job will still be about human connection and empathy, but also deep problem solving. Reps will have more time to provide the excellent customer service they are striving for.

We’re also going to see the creation of new customer support roles. The customer support experience that blends AI and human reps needs to be designed, orchestrated and analyzed and improved—there is a huge opportunity for new roles to be created here.

From conversation designers, who craft the parameters for how AI interacts with customers, to knowledge managers, who carefully ensure the information that AI pulls from for responses are up-to-date and accurate, to monitors, who analyze the bots’ interactions with customers—support agents will have the chance to expand beyond the rote tasks that make up their everyday work today, and embed deeper into their organization.

Marketing Technology News: Six Customer Ratios Give Marketers a Sixth Sense to Increase Lifetime Value of a Customer

A few common misconceptions about AI you’d like to dispel?

The customer service job will still be about human connection and empathy, but also deep problem solving. Reps will have more time to provide the excellent customer service they are striving for.

To best prepare yourself for the AI future, I would recommend starting to get accustomed to new AI tools being deployed and play with them, to see how they work in response to certain prompts, and understand their limitations and upsides