AI is on the verge of becoming a critical part of business infrastructure. TechRepublic spoke to the CEO of a data science company to understand how to best integrate AI into business plans.
With business applications of artificial intelligence ranging from customer service, to hiring, to marketing, it's clear that AI is a tool that is critical for companies to embrace. But using AI has consequences for business structures, hierarchies, and budgets.
To understand how this new technology will disrupt traditional business models, and to help prepare businesses integrate AI in their plans, TechRepublic spoke with Dave O'Flanagan, CEO and co-founder of Boxever, a data science company based in Ireland.
How are your customers (businesses) using AI?
Our platform allows businesses to use AI on customer data that we deploy in a marketing context. We've also deployed it for the call center for customer experience. In the future we see AI fundamentally changing lots of different functions within these large organizations, whether it's how they price products, getting smarter about what the right price for a customer is, or how do they detect fraud patterns from other customers.
How does Boxever use machine learning in understanding data?
We've developed a customer intelligence cloud specifically to help travel retailers get more value from their customer data and connect more effectively in the marketing context or sales context, and to sell to or service their customers more effectively. Essentially, we enable companies such as Emirates, Air New Zealand, Quonset, and others to gather customer data from lots of different silos and then apply machine learning and artificial intelligence to determine the next best action or how to best communicate with the customer at a specific point in time.
That could be sending them a specific email; it could be a mobile push as they pass a shop; it could be understanding that they've lost their bag previously, and the right thing to do is not to sell to the customer but actually to offer them some sort of service recovery. All the time leveraging insights that are surfaced automatically in the data to build better relationships with their customers.
How will companies need to rethink their structures in the face of the AI movement?
On the one hand you could treat it as a huge disruption. We're seeing a huge opportunity to do two things. One, to engineer a cost out of their business, or two, to take advantage of new revenue-generating opportunities. We think that AI enables these organizations to become radically customer-centric.
Both the customer-centricity and the ability to act on the customer, asks a lot of these organizations. What we're seeing is that a lot of organizations are introducing chief digital officers or VPs of Digital who are responsible for the overarching customer experience or the overarching ability to understand that on the data. Whereas traditionally, most organizations are kind of siloed where they've got one function that is responsible email marketing; another function is responsible for the web presence or ecommerce; another function is responsible for loyalty. They've all got their pockets of data; they've all got different strategies or teams of people working on it.
For artificial intelligence to be truly useful and truly holistic, it needs to be connected across all these different functions, and organizations are going to have to think a lot differently about how they want to deploy technology like this to be able to take advantage of it. Ultimately, most organizations today aren't really structured to take advantage of being truly customer-centric and having the ability to act on that understanding with algorithms or insights or machine learning and so forth.
If your company has not planned for this, how do you adjust and weave it into your business model?
When people talk about AI, it sounds big, and it sounds hard. For us, one of the big things that we do with our customers is we help them understand where is the right place to deploy this initially.
What we see is that, let's gather the data from lots of different places, but let's focus on making email smarter. What can we learn about customers? When they open the email, what content should be presented to the customer in the email? How they interact with the email. Then the email gets smarter based on artificial intelligence. That's the way to deploy it into a department that allows the organization to start to build trust, because there's a lot of skepticism about handing over the keys to the kingdom on a customer marketing plan or a campaign to something that's self-learning and automated. We're seeing this a lot.
At the start, Boxever had a black box approach, pure machine learning, where we learned what the right next action was for the customer, and then we executed in an automated fashion. The challenge with that is that a lot of organizations are still getting to grips with one, automation, or two, these self-learning systems.
What we had to do was introduce a lot of controls on rules to be able to allow organizations to treat the output or deploy their own strategies themselves, and then put their strategies in competition or alongside black box or AI strategies to be able to get comfortable with the concept of a machine making decisions about what to send to a customer or what kind of information to present to a customer.
What's the best way for a company to deploy AI themselves?
Start small. If we talk about marketing, pick a very specific area that is low risk/low impact, deploy it, see the value, the return, the uplift, and gradually roll it out over the organization.
The challenge is that AI is typically deployed by "forward thinking" people today, but that won't be the case forever. They've got to help build trust and build credibility and build ultimately ROI to show that this technology, this approach is worth deploying, because there's real value here.
How do you see AI assistants as being important to businesses?
Another area we're seeing a lot of uptake is in the contact center. I think a lot of the mundane interactions can be serviced through automated messaging and AI, but an interesting area that we're focused on is AI assistants. It can help suggest how to deliver that conversation; what interests this person, what kind of products they'd be interested in, maybe what the right thing to say is, and basically making the agents far more engaging, having far more information at hand, and making them smarter.
We see AI not just about pure play automation but actually being able to assist or enhance for my staff to be more engaging and more useful to the end consumer.
AI is a smarter way to automate and change an interaction with the customer.
Why are you skeptical about bots?
In talking about AI and smart bots, what's lost is: how do we measure if it's working? How do I understand if it's truly enhancing or improving things?
When we we start, we're very clear about how we're going to measure it. We try to start small and focus on a specific challenge where we believe we can deploy it and generate results quickly. It's important that we can understand what the business objective is: improving customers' experience, driving out costs, improving conversion? How do we measure it effectively? We must measure and attribute, testing an AI strategy against either nothing or something that was coded up by an analyst.
These are the things that are kind of overlooked when you talk about deploying AI into an organization. It needs a lot of thought; it needs a framework or a platform to be able to understand that it's working and be able to prove it to the company. That's one thing that we've spent a lot of time solving as part of our journey on this process, where we felt we built an AI, a machine learning platform that was able to recommend offers or products to customers.
Does it evolve? It was clear that for it to be successful, we needed to show that it could be tested and deployed in a way that's better than the current system. Sometimes you've got to put in specific rules that mean that maybe AI doesn't work in every context. It's applicable in some but not applicable in others.
Can you tell me when AI shouldn't be applicable? How can you decide where humans should take over?
If you don't have enough data or volume to learn from, you can't train a system and a system can't learn. For example, maybe there's one flight a week from Dublin to New York—a specific route that's bought, and it's low volume—the airline only sells a few thousand or 20,000 of those a year. So in order to learn why people buy that route, there's just not enough data.
In those cases, it makes sense to analyze the data to understand what will be common sense, what will be best practice, and then code that up using rules or something more descriptive. There's lots and lots of cases where you've got to be able to build rules for the edge cases where products aren't available.
That's why I think AI today works well in advertising, where you're trying to find the right ad to show. If you're showing millions of ads, you'll get the sense of what people click on or not. It works well in a call center at volume. It works well on websites. It will work well in IoT where you've got lots of devices sending out lots of different pieces of information at high volume, and you can start to make correlations that ultimately can be surfaced by machines or algorithms.
What new positions will be created?
Certainly we see a huge transition towards chief digital officer. This is an absolutely new role in the C-Suite where they have more IT capability, and they typically have responsibilities for the customer and potentially marketing. For me, the first disruption is our customers adding this new digital capability in their business, and that's naturally where this AI function or AI approach sits.
How will marketing be affected?
What does it do? I think a lot of budget and IP capability is being transitioned into marketing. For some of our customers they're actually moving IP capability out from under the CIO and into the marketing function, so the marketer can do more themselves without having to rely on IT and IT's other priorities to get things done. As that continues, the marketer becomes more powerful with their IT capability; it's a natural evolution that they become the de facto chief digital officer.
I think that's tremendously disruptive. The other key thing is that we're seeing this type of language coming from CEOs of these organizations. They're talking about data. They're talking about analytics. They're talking about how analytics can help their business. In that context, advanced analytics, machine learning, and AI, is key to being able to help them further their business and protect them from future disruption.
How will budgets need to adjust?
We see budgets shifting from IT to marketing. It's still IT—it's just IT in the marketing context. There's a bigger thing happening around chief digital officers, which it could be a more commercial CIO or a more technical CMO, but it's happening. These two things are being sponsored and committed to organizationally by CEOs that are more future thinking and understand that advanced analytics and automated analytics are going to play an even bigger role as their business evolves.
- How to use AI to automatically schedule your appointments with x.ai (TechRepublic)
- 7 trends for artificial intelligence in 2016: 'Like 2015 on steroids'(TechRepublic)
- The original robot "butler": What Zuckerberg can learn from Carnegie Mellon's HERB (TechRepublic)