Building an intelligent enterprise: Four experts offer hard-won advice
At the MIT Sloan CIO Symposium, four IT executives shared lessons they've learned from building an intelligent enterprise and paid homage to the tech that helped them do it.
According to panelists at last week's MIT Sloan CIO Symposium, the ultimate goal is to make every enterprise an...
Continue Reading This Article
Enjoy this article as well as all of our content, including E-Guides, news, tips and more.
"intelligent enterprise." But what is an intelligent enterprise? The title of session provided a clue, associating the term intelligent enterprise with the use that of AI, machine learning, mobility and cloud services.
Hugh Owen, senior vice president of product marketing at business intelligence software provider MicroStrategy, had a more precise definition of what makes enterprises smart.
"An intelligent enterprise is one that uses information to execute on its strategies more effectively," he said. "Every single person in the organization is armed with information, whether it's being given to them because it's aggregated or whether it's AI-driven and has automatically found its way to them. They're doing their jobs better because of access to that data."
Throughout the session, panelists detailed challenges, successes and uses cases they experienced while transforming their organizations into an intelligent enterprise. Here, they share some lessons learned.
What's something you wish somebody would have told you before embarking on your intelligent enterprise journey? Any lessons learned that you want to share?
Stephen Taylor, global head of analytics, reporting, integration and software engineering at Devon Energy, on the dangers of aiming big and the virtue of data enrichment:
I think the [lessons learned] are to start small and start task-based and apply technology for a particular task and then chain a bunch of tasks together. A couple of our projects we had were really large and really ambitious. I think early on ours were too large even at a process level. For example, if you say 'I want to automate the classification of invoices when they come in,' that can be an entire process.
Instead, think about machine learning and determining the classification of an invoice as a step in the process. Try to figure out how to automate that function end to end and get something deployed in production and build more of an Agile approach over time. I wish we would have taken a little bit different approach.
The other one I would add is data quality and data enrichment should be something that you build into the upfront on every project. I think everybody knows data quality -- there are multiple variants to that. But data enrichment is something else. A lot of the data we get, for example, comes in from a particular piece of equipment and we'll have the equipment ID and the values, but there's other information that we can [capture and use]. By using that equipment, we can know what field location it's in, who the operator is, what the operator's superintendent's name was, et cetera. If you can enrich that data then it becomes much more useful in a lot of your downstream systems, if you capture that while it's happening.
MicroStrategy's Hugh Owen on avoiding "crap AI" and making "beautiful" user interfaces:
If the goal is building an intelligent enterprise, the data quality is critical. If you put AI on crap data you're going to get crap AI, so being able to have good quality data and data that you can trust is critical.
The one other piece that I would emphasize is the end-user experience. You could ruin all of the investment we've spoken about if the users are not going to touch it because it's ugly or it's awkward and hard to learn. You need to surface all of this value in beautiful, easy-to-use and enjoyable-to-use applications. That's assuming humans still need to touch these things and it's not all automated work -- which I think is still going to be true. I would say don't underestimate the importance of taking this value and surfacing it to people in interfaces where you've spent just as much time and concern and energy as you have in the plumbing.
Taylor on the right way to ensure automated processes improve upon manual work:
There's one other pretty important lesson learned for us and that is, anytime you implement more intelligent automation or intelligent tasks that are intended to reduce your human labor workforce, a lot of people say humans today are doing a really bad, crappy process. So in addition to automating that activity, they actually want to clean up the process itself. One of the things that we've found has really helped our success is automating the crappy process, taking the people out and then going back and changing the process. That's because if the people aren't there there's not as much resistance. We've found that taking that process, automating it and then going in and tuning it is a lot easier to do than trying to tune it with all the people and their desires to make the process work a particular way.
Alston Ghafourifar, CEO and co-founder of Entefy, Inc., on the money- and face-saving use of experts:
Alston GhafourifarCEO and co-founder, Entefy Inc.
Following what you were just saying, I think the biggest lesson we learned -- and I think it's true for anybody trying to embark in this -- is an intelligent enterprise is too big of a transformation to actually do alone and to do entirely internal. It's one of those things where it just takes too much time, so you have to partner. For us, this is our entire business and it took us four years to build an underlying multi-modal platform that could extend across our company's sets of use cases. Because you also don't want to digitally transform or intelligently transform in just a narrow set of slivers. The whole idea is to do it across the stack both horizontally and vertically.
Most teams in the world are not well-equipped to actually make that transition. It's a set of expertise that not everybody has, and it's also a set of priorities that the organization doesn't necessarily have. If you are not an AI service or product company, that's not your core business. Therefore, there's naturally going to be more resistance -- whether it's intentional or not. You're not going to have the same kind of support ecosystem. But if you partner up with a company whose sole job is to do that and has already spent multiple years on that, you can jumpstart that process, saving -- in some organizations -- hundreds of millions of dollars and at least a few years' worth of development.
Michael Woods, vice president of information technology at CDM Smith, on the wisdom of a multi-cloud strategy:
I think number one would be diversification. Having a centralized cloud -- if you want to call it that -- or a single data center or trying to do everything and put all your eggs in one basket is probably the worst thing you can do. By giving yourself access to as many different cloud systems as you can and diversifying, you're able to access things the minute that something comes around. You've got companies like Entefy and MicroStrategy that provide services. If you can't tap into those immediately you're going to lose out.