We’re all familiar by now, for better or worse, with the idea of the self-driving car, but what about the self-driving enterprise?
That’s the top line pitch from AI start-up Aera Technology, whose CEO Fred Laluyaux argues that AI systems will fundamentally revolutionize the way that executive intelligence is distributed across enterprise organizations, with an inevitable impact for information workers.
All of this means that now is a good time to re-examine some basic assumptions, he suggests, tapping into the self-driving car analogy as a useful jumping off point in his thesis:
What is the difference between a modern car and a Ford Model-T from 100 years ago? The core definition has not changed. Sure, they’ve become faster, more efficient, more comfortable. But they still have four wheels, a drive train, a motor, etc. The core components that are used in the car are identical.
What has changed is the operating system - and this case he’s referring to you and me, the driver.
For more than 100 years cars have used a very unique and adaptable operating system… Humans. We are the central piece of intelligence. We’re an operating system. We learn and we combine the use of our brain to make decisions, our nervous systems, as well as our muscles to enact these decisions. And while so much else has changed in the last 100 years, today for the first time we are in the process of digitizing this operating system.
With the emergence of self-driving automative tech, things start to evolve for the car, posits Laluyaux:
First, it’s connected outside and in. It’s connected to all the events that are happening inside the car. It knows the speed of the car, the tire pressure, how much fuel is left and all the other important indicators. It’s also outside-in: it receives and processes GPS information, radar, sensors, etc. And all that data is processed in real-time. It’s never asleep at the wheel - it’s always on.
The car is autonomous. The ‘brain of the car’ has the ability to take actions. It can inform the driver when something needs to be decided. “There is traffic on your usual route. You should consider an alternate route.” It can also physically take action. It can break, accelerate, switch lanes, park itself.
COS is coming
So far, so Elon Musk. How does this apply to the corporate enterprise? Laluyaux’s argument is that having a Cognitive Operating System (COS) for the organization becomes a game-changer in how that organization is managed and run:
Think about an operating system for a company that is connected to every bit or data that influences it or is generated by it. It’s a system that has been taught how to think, and to reason through most problems. It can learn, it gets better at problem solving over time. And a system that can orchestrate to the nervous systems of a company, its transactional systems?
It’s connected to events that are happening inside of your organization: sales orders, production, purchasing, etc. And it’s aware of outside influences: customers, suppliers weather, regulations, commodity prices, etc. It’s continuously monitoring the business and knows about risk and opportunities in real-time. It’s thinking and can make recommendations based on embedded AI and machine-learning algorithms. It can act autonomously. For example, matching supply and demand in real-time, ship ping orders, rebalancing inventory or adjusting the forecast.
Essentially what’s being pitched here is a variant on the idea of the Digital Twin, one that’s taking on a lot of the work that executives and decision-makers do themselves today, often with some difficulty because of a legacy of data system hell. Laluyaux states:
The reality of most companies around the world is that they live in data and system hell. Forty years of investment in transactional systems and they have hundreds of systems that don’t talk to each other. Decisions are made retrospectively on incomplete, inaccurate and old data wrangled from siloed systems. They often lack the basic cross-business visibility need address core challenges. They have optimized as much as they could on this broken foundation. Billions of dollars invested and the are basically hitting brick wall. For individuals, this is super frustrating. Looking for a piece of information is like looking into a box of Lego for one specific piece, but every few minutes the entire box gets shaken up and things change.
Taking a COS approach begins with harmonzation of systems and taking a leaf out of the exemplar provided by internet-scale companies, advises Laluyaux. He points to Google’s approach to search as a case in point:
Just like Google does not force companies to develop a site with strict parameters in order to be indexed, so to is it critical that your Cognitive Operating System be able to read your data and lakes as they are. You don’t have to start from scratch. The system has to work with what you have in place, crawling if effectively and harmonizing it all together. And just like Google, it will then bring you images of the data picture as frequently as needed, and then let you dive into the data - websites in Google’s case - in real-time when you need to.
Once data is harmonized and flowing in real-time between all the needed areas, algorithms can build on the understanding of end-to-end performance, by layering predictive and prescriptive analytics that find opportunities to improve the financial and operational performance or identify and mitigate risk.
The goal is to get to the stage where a business uses data to have a harmonized, real-time view of operations, end-to-end, with data from internal systems complemented by that from external sources, like weather, regulatory and customer data.
The knock-on result of all this should be standardized metrics on which corporate decisions, at all levels of the organization, can be based. Operational decisions can be automated, while the harder strategic decisions that need human intervention are augmented.
And the COS will grow and expand as it learns from its own experiences, evolving from initially making suggestions to its human owners that can be accepted or rejected. But from this supervised learning process, the COS will learn to make predictions and come to its own conclusions about what decision will be made by the human beings to its recommendations.
It’s all about memory, argues Laluyaux:
There is no intelligence without memory…Each recommendation and prediction is aligned across organizations, across departments. This means memory is being effectively shared across a company - shared, and retained! The company gets smarter. Companies have a live memory of all the decisions that are made. They have a digital workforce that can actually drive the business. Now, finally, we’re ready to talk about self-driving.
Breaking down the pyramid
It’s a big picture vision and clearly one that has implications for the traditional pyramid organizational model. But maybe it’s about time that changed, suggests Laluyaux:
For hundreds of years the structure of enterprises has not changed - a nice big pyramid where we are programmed to do our jobs, fit into a process- and try to move up one level every so often? The way we learn - or get educated- has barely evolved. The way we are compensated has barely evolved. Technology has is predominant in companies as well, of course, just like with cars, but the operating system has not changed.
And it should have, he adds:
We have access to a lot more data and tools that before. We communicate faster and more often. Transactional systems are now handling a lot of the manual work and as a result they have enabled the globalisation of our companies and of our economy. But they way we decide has really not changed.
Companies will change to become built around ‘tribes’, says Laluyaux:
We believe that instead of this old pyramid scheme, this hierarchical structure, people will be leveraged in an efficient and dynamic network of intelligence. Where communication flows swiftly, people will have more success and loyalty tied to their community or tribe, and not to one or a couple of organizations. The structure of intelligence will move from a pyramid towards a network. It won’t be such a clear-cut path towards career advancement, not a ‘be a good soldier and all will be clear’ world. It will be a world where intelligent agents - humans and machines, are distributing themselves through the job and work network.
Each role, each person, has a Digital Twin. It means each executive gets personalized recommendations and feedback about their decisions. We will have to learn how to work with these Digital Twins, just like workers learned who to work with robots in a manufacturing plant.
Our ability to ask questions and anticipate events will be increasingly more valuable than our ability to execute tasks and react to events. The EQ will win over the IQ. Our ability to unlearn our old ways of doing business will be critical to our ability to learn and adapt to this new era. I think that to apprehend this new era, we will have to approach it with a fundamentally different mindset - one of growth and change, and constant learning.
It’s all about changing to meet the new demands of the competitive landscape in the end, says Laluyaux:
It is no longer the big that eat the small; now it’s the fast that eat the slow. This is why self-driving companies will win!
An interesting thesis from a company that’s generating a lot of buzz in Silicon Valley. Clearly the self-driving car analogy immediately invites cynical comments when another autonomous vehicle hits the wall, but the comparison is still apt. Just as human beings are going to have to adjust their thinking and expectations as self-driving taxis and the like enter the mainstream, so too there’s a need to re-think the approach to enterprise information sharing and decision-making. There’s a lot to work out of course, but Laluyaux’s premise provides a lot of food for thought.