Interview with Or Shani, Founder and CEO at Albert

The complexity and deluge of data that this new normal generates is daunting for humans but generates an ideal environment for a machine to thrive.

Know My Company

Tell us about your journey into Artificial Intelligence? What made you launch an AI platform for marketing?  

I began developing Albert in 2010, after having worked my way up from an engineer to a CMO in various marketing companies. I realized early in my career that even newer digital marketing techniques were still far too manual. Considering the technological innovations in other fields, I suspected there had to be a way for technology to handle the technical and data aspects of digital marketing campaigns, while humans handled the creative and strategic components. That better way was Artificial Intelligence.

I put together a team of data scientists, engineers, and mathematicians, and we got to work reverse engineering the logic and intuition of marketers to intimately understand their thought processes at every step of a multi-channel digital campaign.

After six-plus years, we had developed a fully automated, digital m***************** that doesn’t just replicate what human marketers do, but works at a scale not previously possible by even the largest marketing teams.

I assumed all upfront financial risk, fully self-funding the company for the first 5 years (though our initial platform reached profitability almost immediately). We were offered capital from VCs, private equity firms, and other investors, but declined all of them to maintain the integrity of the company and technology.

In 2015, we took the company public, raising $42 million on the London Stock Exchange to ensure Albert was first to market in the US ahead of other business AI platforms that launched soon after us.

Over the years, I faced pushback from the very people who stood to gain from Albert—marketers who feared AI would take away their control or even their jobs. With constant education, we’ve been able to shift this conversation dramatically.

What are the foundational tenets of your AI-powered platform?

Before we broke ground on Albert back in 2010, we knew we had to first establish our unique philosophy around marketing and advertising. We didn’t want to build a platform that simply enabled and scaled the status quo; we needed to predict the future to understand what the landscape would look like when we launched (2016) and well after that. We established the following four guiding beliefs, which continue to guide development and push the marketing industry forward to this day:

1. Brands will become retailers. Eight years ago it was already clear that brands would need to begin creating direct relationships with consumers and customers, rather than let retailers dictate the terms of engagement or otherwise stand between brand and buyer. This belief is not specific to retail brands; we’re seeing now that financial institutions, telecommunications companies, and other traditional industries have also had to master retail basics to keep up.
2. Brands must master the customer journey. Taking on the role of retailer—whether from a direct relationship standpoint and/or a transactional one—means keeping up with and communicating with consumers wherever they are. As the number of consumer touchpoints continues to expand massively across devices and channels, online and offline, brands need more effective ways to meet customers where they are. The complexity and deluge of data that this new normal generates is daunting for humans but generates an ideal environment for a machine to thrive.
3. Digital transformation isn’t enough for enterprises to keep up; they need AI transformation. As digital transformation sweeps and transforms every industry, disruptive new players emerge to solve existing problems and offer a new customer experience that major legacy companies simply can’t. This creates an urgent need for established companies and enterprises to reinvent themselves and find more effective, efficient ways to engage customers
4. AIs are a new kind of colleague. For AI transformation to be successful, there must be a symbiosis between man and machine. Humans and machines are partners, recognizing what each does best and using their individual strengths to heighten overall levels of performance. Humans, for example, are needed to guide strategy, dream up creative, and augment AI applications. By doing this, they’re enabling these systems to function as true collaborative partners. Machines, in turn, are expanding human capabilities—providing marketers with superhuman capabilities, like the ability to process and analyze huge amounts of data in real time.

Not all AI is autonomous, but Albert is. What makes an AI autonomous?

There are two broad categories of AI: AI that specializes in crunching data and offering insights, and AI that does both of these things and then acts on its own insights in pursuit of a predefined goal.  The latter is autonomous AI, which is the category Albert lives in.

Albert not only collects, interprets and predicts the next-best action in every channel for every audience and ad creative, he then executes and optimizes against his own conclusions in real time.

This creates a clear division of responsibilities, as well as opportunities for collaboration, between humans and machine: Humans set up initial parameters and business rules against which AI can execute. Machine focuses on data and technology. Albert does this 24×7, 365 days a year, leaving his human colleagues to focus on high-level creative tasks, customer experience and brand while Albert.

Define your Ideal Customer.

Consumer-facing businesses looking to drive clear and quantifiable actions: purchases, sign ups, bookings, even phone calls. Often these companies are in retail, apparel, CPG, travel or financial services with a typical ad spend of over $500K per year in digital media. We work with companies located all over the world.

What makes AI and machine learning algorithms so attractive, yet elusive to most B2C businesses?

Algorithms themselves are fundamentally undifferentiated. They are available to anyone on the web and interesting because they have the potential to help solve complex problems. The challenge for any business is how to apply the correct algorithm to a specific problem. For example, in digital advertising, campaign managers often face the question: if one channel or audience or ad is performing better than another, how do I know how much money to move to drive optimal performance?

If a channel is 2x as effective, do I spend 2x as much? Not necessarily.

This I where a predictive model comes in.

The optimum point for spend performance is not linear, it is usually logarithmic. But it is different in every situation.

Predicting the impact of an investment change on performance must be done with all kinds of regressions. It is a very complex problem, but a machine can do it better than any human. Machine learning, however, is not the interesting or elusive thing here. It’s arranging data for machine learning that acts as the barrier to entry, as it requires enormous effort. And considering that this is just one aspect of a digital campaign, the real challenge can be the complexity of many algorithms running alongside each other and determining which takes precedence, when and how.

Which markets and geographies would you consider as AI-friendly? 

Like all digital technologies, AI is borderless.

It does not see geographic boundaries; it simply sees variables, which may differ from region to region, and possibly drive different strategies for success. That said, there are optimal circumstances for AI. AI is taking off in regions that have already optimized work as well as possible using existing technologies.

For instance, a society that is still struggling with consistent access to electricity is not going to jump from solving that problem to AI. According to the World Economic Forum, AI is ushering in the fourth industrial revolution, so it’s logical to say that the barrier to entry is less about geography and more so related to which “industrial revolution” a market currently operates in.

What makes Albert a go-to tool for the Marketing teams? How could it fit into a modern CMO’s tech stack?

Scaled performance without scaled operations, performance-based pricing, increased ROAS and ROI, and the ability to identify high-value audiences brands don’t even know they had. Albert doesn’t just manage all the mind-numbing, manual tasks required to manage and optimize multi-channel digital marketing campaigns; he does them at a scale that’s not possible by even the largest human teams.

The marketer provides the goals and creative elements, and Albert takes it from there. He works 24×7 comprehending the massive and continuous deluge of data, channels and devices, creating real time micro segments and campaigns that engage across the customer journey in a way only a machine can. As one of our clients puts it, “What I like about Albert is that he doesn’t break up with his girlfriend and slack off the next day. AI doesn’t need to take a vacation. It just works hard.”

Albert is like having a digital colleague, he will ask for creative and strategic help when he needs it, and let the marketer know about untapped opportunities and insights he discovers. Ultimately, he delivers true cross-channel marketing, tying brands’ digital siloes together and orchestrating results-driven, holistic, people-based digital campaigns while optimizing efficiency in digital ad spend.

How do you make AI deliver economic benefits as well as social goodwill?

Albert magnifies any campaign that is put in front of him. For brands that have economic goals, there is a direct relationship to magnified revenue. For brands or entities that run campaigns for social good, Albert can similarly magnify awareness and responses. An everyday way that Albert promotes social goodwill is by not pissing people off with irrelevant marketing. (Sorry, had to go there!)

But there have been cases where our clients have used Albert for true social good, such as when a Texas-based client used Albert to get the word out about access to shelter during Hurricane Harvey or a technology company that works with “locked-in” patients used Albert to educate patients about new options for managing their way through it.

Worth noting is that Albert lacks human bias, which lets him transcend long-held (and often limiting) beliefs about people in general, and more specifically, a brand’s customers.

How do deep learning algorithms and natural language processing technologies converge at Albert?

We’ve developed a proprietary, patent-pending intelligent machine that leverages predictive analytics to gain insights from massive, disparate data sets, and then use deep learning technology to act effortlessly on unpredictable situations that have previously forced human marketers to make risky bets with limited information.

The predictive analytics and deep learning technologies come together with natural language processing to comprise machine learning—or expert systems based on our unique rules and methodologies.

All of this is what allows Albert to observe data patterns—as well as more abstract stimuli—and draw conclusions from them. He then autonomously adjusts his behavior without the need for additional programming or decision making by the marketer.

Tell us about your AI research programs and the most outstanding digital campaign at Albert?

We are currently working on an AI Operator program, dedicated to converting marketing experts within the enterprise into everyday AI experts who will act as conduits between AI and human colleagues.

One part AI whisperer, another part operations professional, the AI operator’s sole purpose is to strengthen AI in places where human intervention is required: communicating what it’s doing and learning, understanding business goals, and offering its thoughts in the form of insights (rather than raw variables considered in the decision-making process).

These AI experts will act as an interface between teams, pushing projects forward and putting the AI’s insights to work to help people do their jobs. They will set parameters and guidelines, speak “robot,” and translate what it has to say back into “human.”

In terms of outstanding digital campaigns, we have been able to achieve replicable success across clients and client types. Some of our most successful campaigns to date have been for Harley-Davidson, Cosabella, Dole Asia, Natori, Gallery Furniture

, and RedBalloon.

What are the major challenges for AI technology companies in making it more accessible to local communities? How do you overcome these challenges?

One of our biggest challenges is analyzing a new client’s business and translating it into something the machine can understand in binary software. We often refer to this as “learning to speak robot.” Every client undergoes an AI transformation on their journey to adopting Albert. We take a deep dive into their business in detail to configure Albert’s definitions. This is followed by an initial learning phase where Albert, like any new employee, learns the ropes.

His human colleagues also go through a process where they learn to trust him as he tests and learns. Once we have successfully applied business logic to the machine, again, just like a new employee, Albert can run and add value on his own, only asking for help in the form of new creative, or for decisions about his recommendations or opportunities he has discovered.

The Crystal Gaze

 What technologies within AI and computing are you interested in?

1. Self driving car
2. Google deep mind – because they are pursuing general AI
3. Watson medical
4. MIT Computer Science & Artificial Intelligence Laboratory

As an AI leader, what industries you think would be fastest to adopting AI/ML with smooth efficiency? 

1. Marketing
2. Medicine
3. Financial Services
4. Weather
5. Automobiles

Tag the one person in the industry whose answers to these questions you would love to read: 

Daniela Rus, Director MIT Computer Science & Artificial Intelligence Laboratory (CSAIL)

Thank you, Or! That was fun and hope to see you back on AiThority soon.