In the early 1990s, established businesses were blindsided by the disruptive force of the Internet. Even those who invested heavily in IT infrastructure soon found that they had difficulty competing with nimble new upstarts who could deploy web based applications at blinding speed.
It was with this in mind that IBM launched its e-business initiative in 1996. Although considered by many at the time to be an aging dinosaur itself -- it was close to bankruptcy three years earlier -- it was able to leverage its expertise in legacy systems to transform business processes and prepare its customers for the Internet age.
Today a similar tidal wave of disruption is sweeping through the corporate world -- artificial intelligence -- and, once again, IBM sees a big opportunity. Much like in the early days of the Internet, there is enormous value to be unlocked, but legacy systems are not set up to leverage it. It's not enough to simply apply algorithms, you need to prepare your business for a new age.
Building An AI Application vs. Building An AI Enterprise
These days, everybody can access capabilities that would have seemed like science fiction a decade ago. Using so-called "no-code" platforms like Mendix, Zudy and Quick Base, even those with little or no technical training can pull powerful tools from places like Amazon's AWS, Microsoft's Azure and IBM's Bluemix.
To understand how it works, think about a retail brokerage that wants to create an app that allows its customers to access their account information using only their voice. Employing existing infrastructure to access the account, they can pull voice recognition technology in through API's and make it available on mobile phones or in a "smart speaker" like Amazon's Alexa or Google Home.
With just a few more clicks, the application can be connected to data API's to access market information so that customers can find out how the the market did today. Once again, the technology is now simple enough that an ordinary marketing manager can build the app with little help from the IT staff.
However, from there things get dicey. If, for example, a customer wants to open a new account from the same app or access analyst reports, different systems come into play and these are often incompatible with each other. So if you want to truly leverage the data in your enterprise, building applications is not enough. You need to become AI ready.
Making Your Enterprise AI Ready
A typical enterprise today is sitting on a wealth of data that it has been collecting for decades, but most of it is inaccessible because it was collected for a specific purpose. For example, a retail brokerage might have different systems for customer accounts, trading, market analysis and so on.
Typically, each of these were built at different times with different technologies, so most organizations have valuable data spread across a variety of platforms and formats. Some of it is in traditional databases, some in hadoop clusters and some strewn across the enterprise in Excel worksheets. It's all there somewhere, but getting to it can be an enormous problem.
Another issue is with data governance. Because data was collected at different times for different purposes, it is often stored in different formats. There is also often significant duplication. A customer's account in a retail brokerage, for example, might have different versions for customer service, billing, marketing and so on. Some of these might also have collected different information, like address changes and maiden names over the years.
"To prepare for AI, you first need to build a unified and integrated data environment," Rob Thomas, General Manager at IBM Analytics told me. "Much like in the previous transformation to a web powered enterprise, there's no simple recipe. Every enterprise is different and needs to build its own data culture. But if you want to do AI at scale, you need to make your data AI ready."
Data Driven Business Models
Most businesses start thinking about AI in terms of their current business model. Using basic AI applications to automate customer service, for example, can save time and improve customer satisfaction. If a customer can check on their account merely by asking, "how did my portfolio do today?" while they're stuck in traffic, that's far more efficient than having to call a broker.
However, Thomas sees far more potential in using AI to transform business models. To understand how, imagine a typical meeting with a broker 20 or 30 years ago. A good broker would spend some time analyzing the account history, reviewing the customer's investment objectives and exploring options to improve performance.
The Internet streamlined this process, allowing an investment advisor to access information much more quickly. So in the process of a short phone call, she could revise her analysis based on changes in customer needs. For example, if customer informed her broker that investment objectives shifted from saving for college to saving for retirement, a new analysis could be produced in a few clicks.
With an AI powered enterprise, its the analytics itself that can be transformed. So if a customer wants to know how her portfolio could be affected by a rise in interest rates, todays systems can access not only databases, but unstructured data like analyst reports and predictive algorithms. What's really exciting is the potential for customers to be able to query the systems themselves, using their mobile phone or smart speaker.
Take the story to its logical conclusion and it becomes clear that we're not only talking about a change in data or analytics, but a very different kind of business. Once customers can access the systems themselves, brokers need to provide a very different kind of service. Instead of merely providing answers, they will need to learn how to coach their customers how to ask the system better questions.
Shifting The Data Mindset
Before the Internet, most large enterprises acted as gatekeepers for information. They knew much more about the performance and pricing of their products than customers and could leverage those asymmetries to juice profits. The web changed all that and businesses needed to adapt to survive.
Today, we're in the midst of a similar transformation, but instead of information its expertise itself that is becoming democratized. Powerful analytics, combined with artificial intelligence applications that not only recognize voice and visual inputs, but indeed learn themselves, are making it possible for machines to perform tasks that were once considered intensely human.
IBM's Thomas believes that this means that enterprises need to change the way they see data. "Traditionally data has been all about reporting on historical performance -- 'tell me what happened,'" he says. "But today data needs to be more forward looking, we to be able need to use it to predict, anticipate and advise.
That means that once again businesses will need to adapt. Just as the web made it essential for organizations to open up their data, the AI era will make it necessary for enterprises to share their expertise. But before that can happen, we first must make our data ready for it.