If you’ve been adopting a wait-and-see attitude about AI, Justin Reilly suggests you get started now. “The rate of adoption for this technology is starting to pick up, and it’s important to understand how it might play a part in your business today,” he told the audience at ATB’s 2023 Business Summit. Take ChatGPT. The natural language processing tool reached 1 million users in just five days, and 100 million in two months. By contrast, TikTok needed nine months to hit the same milestone.

Reilly is the CEO of Wavelo, software that powers telecom businesses, and former Head of Product at Verizon. With his experience in digital transformation and machine learning, he knows that bringing AI to your business requires courage, audacity and a lot of work. To help with your business’s AI implementation, here are the five things he recommends you consider.

Data architecture

The first question to ask yourself is, What’s your business’s most valuable data? Once you have that answer, the second question is, Is it easily accessible? By that, Reilly means you could get on your phone right now and retrieve that data in real time. “You need data in motion to get the most value out of this era,” he noted. “If it isn’t easily accessible, you can’t train on it.”

Those are two daunting questions, but Reilly suggests thinking about what your “minimum viable data” is—the three to five pieces of data on your business that add maximum value. You can always iterate later, but those biggest pieces of information are a good place to start. They won’t overwhelm your team or your partnerships, and you’ll quickly see value.

Talent diversity

You know that talent diversity can be a competitive advantage, but how you think about your talent stack will probably need to change. As Reilly said, what got you here won’t get you there. “It is almost certainly true that the person you need to lead this inside of your organization doesn’t exist in your organization.”

If you intend to build an internal team focused on machine learning, for instance, you’ll need a machine learning engineer and a researcher. However, it’s more likely you’ll hire someone to lead AI partnerships and integrations. This is the person who will vet your AI partnerships and ensure the integrations deliver the most value.

Customer experience

Since the dawn of the internet, the evolution of customer interfaces has been tremendous, from static sites on a desktop computer to that supercomputer in your pocket. Now, the interface is agnostic. It could be a website, a mobile app, a voice interface or a truck. Simplify things by considering how your customers might interact differently if they could truly have a conversation with your business with limited human interaction.

But this isn’t about eliminating the human element. Think augmentation over automation. “Take it from someone who had lots and lots of capital inside the world’s largest telecom at my disposal: it’s really hard to automate end-to-end processes that have had humans in them for a long time,” Reilly said. “It’s hard to get it just to the level of goodness when the humans were involved, and the bang for the buck tends to not play out.” Full automation may be possible in the future but, for now, most of your value will come from augmentation—helping your human workforce become more efficient. Make augmentation your core strategy.

Dan Semmens, SVP, Head of Data & AI at ATB, agrees. Of the over 200,000 AI-assisted client interactions ATB delivers every year, most include a human team member. “AI is never going to replace those humanistic characteristics that are so key in how we serve our clients—creativity, empathy and compassion,” he said. “When we think about how we deploy AI tech, it’s very much ‘human and machine.’ It’s in that augmentation where great experiences result.”

Partners

Reilly uses “partners,” not “vendors,” intentionally. “Because, unlike every other era of technology, your relationship with your partner is going to be more intimate than it ever was.” To get authentically cognitive products, the technology needs to continuously learn. That means you must hand over your data to your partners. “You need to really trust them, and they need to be invested in your business.”

When you’re vetting AI partners, he has two “cheat codes”:

• Ask them about their organization’s ratio of machine learning engineers and researchers to marketing and sales people. This will give you an idea of whether they’re building the tech themselves or partnering externally. You’ll understand where your data is going and who’s involved.
• Lean into proofs of concept. If they’re not willing to offer a proof of concept, there’s probably some smoke and mirrors. This is particularly important for medium-sized businesses, which can’t afford to get it wrong and have to pivot six months later. “Lean into proof of concepts and you’ll find your best partners there.”

Policy

Developing a corporate policy on how to use AI technologies and how they interact with your data can feel overwhelming. Reilly recommends keeping it simple: start with your existing policy and extend it to AI. You already care a lot about your and your customers’ data. And you’ve already thought about how to protect it and the ways you’ll use it. Extend that thinking to AI.

For example, as ATB advances our artificial intelligence functionalities, we never lose sight of our core values. “Given that we’re a purpose-driven organization, it’s really important that all AI solutions we undertake align with our ethical principles and brand values,” Semmens said. “Everything we do when it comes to artificial intelligence begins and ends with our clients in mind. We want to ensure we can demonstrate we’re following responsible practices and prioritizing our clients’ well-being.”

The nuance, Reilly noted, is the conversation you have with your customers about how you’re using their data. Be customer-centric. Tell them how you’re using their data and the benefits to them. Maybe it’s better service or an improved product, a discount or the chance to participate in a beta program. “Tell them the why. You might be surprised at how excited some of your customers are when they know they’re getting value on the other side.”

Where to start

Reilly suggests three things you can do to start laying the groundwork for your foray into AI.

Articulate your big rules—and no more than three of them.

One might address whether you want to build your own tech, partner externally or adopt a hybrid approach. Another could be to consider your policy position. In one sentence, articulate what you as the leader of the business want the position to be, then let your team take it from there. Or you could sketch out your operating principles. Having clear operating principles gives your employees “a GPS in that moment” when they need to make a decision but you’re not there to guide them.

Find a low-cost, high-value way to get started.

In other words, find the “least painful” way to get that first view of value. But, at this point, the value isn’t the most important thing. “It’s the cognitive load you’re going to go through to go ‘Oh, all right when I did that, this happened.’” A lot of the patterns you built your business on aren’t the same. Working through that process will spark thoughts on how the technology can fit your business.

Become a user.

“This is not in your roles as business owners. This is in your roles as human beings walking around the Earth.” Reilly suggests starting with OpenAI or Midjourney and playing with them so you understand how they work and how the models come to life.

ATB and AI

ATB is using artificial intelligence to enhance both our team members’ and clients’ experiences. For team members, it’s simplifying complex tasks, helping them make more informed decisions based on data and freeing them up to focus on the value-based work that drives ATB, like cultivating relationships with clients.

ATB clients are benefiting from more personalized and contextualized experiences, improved fraud detection and credit adjudication, and forward-looking insights, such as predicting credit distress. During Alberta’s 2023 forest fires, we even used AI to monitor transaction patterns to help us support clients affected by the fires.

Regardless of the applications, Semmens says the guiding principles remain constant. “We’re focused on enabling experience. We’re using AI toward a value-based outcome—for both our clients and ATB.”

If you want to keep up with developments in AI, Reilly recommends these newsletters:

AI Weekly by VentureBeat

The Algorithm by MIT

OpenAI

AI to try

DoNotPay—A “robot lawyer” and consumer champion that helps you do things like break up with your gym, write a letter to your landlord and navigate the IRS’s phone menu.

51WORLD—A digital twin technology provider supporting smart city planning, building construction and management and infrastructure (check out its Shanghai project).

Box AI—A large language model in Box that can summarize your documents, draft content, provide feedback on your material and much more.

Midjourney—An image generator that creates images from natural language descriptions.

Confirm—A performance platform that powers performance reviews, conducts and categorizes employee engagement surveys and assists with talent management.

Meet the speaker

There’s no better guide to the pitfalls and potential of AI First than Justin Reilly. He’s spent years building businesses around the immense capabilities of Artificial Intelligence. As Verizon’s Head of Product, he spearheaded their multi-billion dollar digital transformation program, which had machine learning at its core. He’s currently the CEO of Wavelo, a division of Tucows that’s revolutionizing the software that powers telecom businesses, where he’s building the modern telecom platform aimed at adapting faster and innovating better for an AI First world.