Integrate.ai and the Obligation to Responsibly Mature AI
AI is big business—stop me if you’ve heard that one before. All around the world, AI startups are flourishing, even at the same time that the biggest companies in the world build their own divisions in the burgeoning field. But many AI firms, both small and large, are finding themselves at a point where they must step back to recognize the true repercussions of their relentless pursuance into data science. Many symptoms of the AI surge are great, resulting in fundamental redesigns of how almost every industry conducts business.
The other side of the AI surge is a bit less positive: The emergence of bias in machine learning or the assumption that AI might be used nefariously by rogue actors. It’s taxing to fully grasp the ramifications that the widespread adoption of AI will bring and many companies simply choose to ignore any ethical or moral implications. That’s not the case for Integrate.ai. With over $50 million CAD in venture-backed funding, the Toronto-based startup, which turns two in just a few days, works with industry-leading partners yet also finds time to focus on the ethics of the business they’re in. Integrate’s model could generally be described as an AI-as-a-service (AIaaS) platform. But Steve Irvine, Integrate’s founder and CEO, doesn’t necessarily like that difficult-to-pronounce acronym.
“I think we’re a little more focused than a general AIaaS platform,” Irvine explains as he sits in Integrate’s fifth-floor boardroom. “The big problem we want to solve is that a lot of traditional companies like banks or retailers want to operate sort of like Amazon, in a very customer-centric way. The challenge is that they can’t recruit the same quality and depth of talent, and they don’t have enough data and focus to do that.”
“What we try to focus on is creating a software platform that would be essential for them to start looking and operating more like an Amazon. I think we’re absolutely offering AI-as-a-service, but I would say our focus is much more on a discrete set of customers and a real problem that we’ve identified.”
If Not “As a Service” Then What?
Nailing down exactly what Integrate does is a bit difficult. The startup works with a wide range of partners (publicly stated partners include Corus, Telus and Scotiabank) so that means the AI solutions the company builds can differ greatly from one client to another. That also means the impact Integrate has with customers differs as well. Sometimes they can help a company improve productivity and revenue by 40 per cent, Other times it’s only one per cent. But those numbers mean different things when working with a 20 person startup or a 6,000 person enterprise.
Integrate’s COO Jason Silver remains focused on one thing: impact. This is a repeating trend that surfaces throughout the company, and it’s an ideal that IAI-ers (as they call themselves) value most above all.
“We’re trying to keep focused on the impact we’re having,” Silver says. “We’re looking for very specific applications. It’s more about us finding the right place to deploy the kind of impact we have.”
“It’s easy to describe what impact is not. We’re not interested in telling businesses how to operate—we’re interested in integrating with businesses and helping them bring change to life.” – Jason Silver
The AI platform that Integrate offers its clients centers around identifying insights and aiding businesses as they move from traditional decision making processes to frameworks guided by machine learning algorithms. This, in turn, allows clients to make use of the essential service AI has become without having to hire a team of data scientists themselves.
“The question was around helping more traditional consumer businesses compete with the biggest companies in the world to start to deliver both better customer satisfaction as well as better profits,” explains Irvine. “What we’ve seen from our growth is that a lot of businesses have struggled with how they bring this stuff into the company. There’s been a lot of hype and buzz in AI, but when you drill down and ask ‘how do I make my business better tomorrow,’ that was a big problem for a lot of traditional businesses.”
One of the most impressive parts of Integrate’s AI service is their ability to safely and securely draw from multiple datasets to help any given company. The startup’s Trusted Signals Exchange system uses deep learning to recognize key insights from different clients and apply those learnings to a new environment, all without breaching privacy. In this sense, clients of Integrate who do not have access to rich datasets themselves can draw from other clients who do and act like a big fish swimming in a small (data) lake.
At the end of it all, Integrate’s business model is based on the simple model of how tech markets have played out over the last two decades. As Irvine explains, at the beginning of the internet, companies would build sites for clients for a lot of money, until big platforms found streamlined ways to do it. Same with apps—companies would build apps for lots of cash until a streamlined platform was made to do it quicker and cheaper. In these early days of AI, firms are still currently delivering deeply-customized AI models to clients.
“Our bet is that AI will move towards a big dominant platform that can deliver these rich capabilities in a much more scalable way that meets the economic requirements of running a great business, but is also easily consumable and integratable into businesses that will need these AI capabilities,” says Irvine.
“The goal for this company from day one has been to create a software company that is powered by AI.” – Steve Irvine
Always be Impacting
Having an idea to create a software company is one thing—leaving a high-paying marketing executive job at Facebook in the Valley to come to Toronto and start a small (at the time) Ai firm is another thing entirely. That’s what Irvine did two years ago when Integrate first opened its doors, and as they say in tech, the rest is future-proof.
Building an AI company in a city and country distinguished by its AI prowess is daunting. Integrate built their foundation by relying on a core loop of feedback as well as five values: Focus on impact; take action; be present; love people; build trust. These commandments have been the guiding vision as Integrate scaled massively over the last 24 months.
“One of the things that I’m most proud of about this place is that our culture is as strong now as it was when we were in the early days of the company,” says Silver. “That’s very hard to do and it’s a huge testament to the people here and how much they care about the work they are doing. The day-to-day and the things you work on change, but the way people show up and the respect they have for each other, that has stayed constant as we’ve grown.”
Culture is a volatile word in the modern technology industry. As Silver notes, it can be defined by ping pong tables, but his definition of culture relies entirely on how someone does their work. In the AI field, markets will shift every quarter, so his ideal culture is ensuring the IAI-ers are empowered and accountable, and that the surrounding team trusts them to get projects done.
Integrate’s definition of culture is guarded by a somewhat-intimidating habit of deep feedback. The company has a living document that always updates exactly what their definition of feedback is, and it is handed to new hires as part of a welcome package. Along with that, each new IAI-er must give and receive feedback in their first two weeks. No, it’s not normal to tell a bunch of newly-introduced coworkers how to do their jobs better, but that sense of feedback drives Integrate.
“If you can build a feedback loop so that it self-reinforces and identify that as a core value, you get people very comfortable with giving and receiving feedback so it becomes a self-regulating environment,” says Silver. “As you scale you get more feedback, people get better, and you grow faster.”
This culture breeds a certain ethic of self-improvement within the company. Each employee has a polaroid of themselves on a vision wall, showcasing their personal and professional goals for the year, along with notes on what kind of help they might need from the team to achieve those goals. And if providing day-one feedback isn’t scary enough, each IAI-er also has to paint a small canvas to hang on the wall as well.
The Moral of the Story
Integrate’s dedication to platforming AI comes with a price that Irvine and the team recognized from day one. If they will be creating AI solutions that interact with millions of consumers, they have an obligation to address one of the field’s greatest problems: bias and stereotyping. As such, Integrate has a dedicated approach to instilling a staunch commitment to ethics and morals in machine learning—so much so that they have published a whitepaper on the topic and Irvine regularly shares perspectives on the issue.
“I don’t know if everyone understands how big of an issue the ethical side of AI is when it comes to understanding fairness and bias,” Irvine warns. “It’s not that robots will take over the world.”
“Instead, it will be using datasets from historical data that a lot of companies have collected, where there will be an inherent bias of the past,” says Irvine. “Without adjusting for that, we risk perpetuating the stereotypes we’ve been trying to progress away from for decades. We’ve got a unique opportunity as an AI startup that has been built from the ground up to focus on security, privacy, and ethics to build without constraint the right things for today’s world.”
In fact, in order to align with Integrate’s culture as a scaling company, there’s really no other way to approach the notion of ethics and fairness in AI. Creating a lasting impact with a company means prepping them for the future, and if that future is created by ambiguous data sets and morally corrupt algorithms, that company will have created the wrong sort of impact, and Integrate’s employees will have gone against their own values.
“There is no other way for us to build this company in line with our values that does not see us investing in the ethics and morality side of AI.” – Jason Silver.
“This company has to be built in a way that adheres to those values,” says Silver. “How can we build trust if we have an algorithm out there that is knowingly unethical? How can we build a company that loves people if it’s not acting in the best interest of those people? You can’t hold a company to different values than the people are held to. Calling it a smart investment in ethics is one way to look at it, but to us, it is the only investment. There is no other way.”
Keep on Scaling
With $50 million in financing and some of North America’s biggest brands as clients under their belts—all before their second birthday—Integrate has a lot of work to do if they want to keep scaling as quickly as they have been. But it all started with a conscious decision from Irvine to move from the Valley back to his native country and invest in the ecosystem there.
“Two years in, I stand behind the notion that Toronto is the best place in the world to build an AI business,” he says. “Everything I’ve seen in these last two years is validation of that. I don’t think there’s a scenario where we could have scaled quicker or attracted better talent. I think we have shown in two years that you can a really big global-leading business out of this country.”
The next step is to take this Toronto company to the world, fuelled by Integrate’s late-2018 $39 million funding round.
“The aspiration is to be a global-minded company. 2019 will be the year we put a stake in the sand and expand into the U.S.,” says Silver. “Looking at putting offices in the U.S. will be based on who our customers are and what they need from us.”
“The vision overall is for us to build a lasting business that sits at the centre of a very large market.”
