Aera CEO Laluyaux - AI, self-driving and why it's time to stop being excited by old enterprise software 'innovation'

After starting work in the enterprise software world some two decades ago, Aera CEO Frederick Laluyaux came to a realisation a few years back:

We were in the rat race where giving more data and more point solutions, that would be sitting on top of transactional systems, was just an unwinnable war. You know, we would be, ‘Here is more data, more tools, go and figure it out’. I realized years back that it was obvious that every three years we'd come up with a better incremental version of the last solution and I got tired of it.

The complaint from users was that this was just “getting us to the wrong number faster”, he adds:

There's really no intelligence in what you do. You're just removing pain, removing pain in the process, killing a spreadsheet with a better tool and so on and so forth. It became very clear to me that the next wave of change in in the enterprise world, was what I call Cognitive Automation. It’s automating how decisions are made and executed. That became possible when for the first time we're able to combine three things that were not combinable before - complexity, transactional level volume and real time, collaboration.

Until now it’s been a case of being able to do two out three, he says:

I can build complex models and I can give you some real time calculation in minutes, but don't ask me to work at a transactional level. If you want to work at a transactional level, I’m going to have to compromise either with the response time or this or that. You’d always end up with this conundrum.

Self-driving

What changed the status quo was the emergence of what Laluyaux calls “Internet-scale technology” in the shape of Google, Facebook, LinkedIn et al:

Those guys came up with a technology, an architecture and a full process that are completely different. So we leveraged that to build what we call the Cognitive Operating System…it’s enabled self-driving.

Self-driving in this instance is applied to the enterprise, although the application of the term to the automotive industry is pertinent:

If you think about the self-driving car, they’ve digitised the operating system of the car, but that’s been a human so far. You need to think about what you’re doing. [In the enterprise] we’re digitising and upgrading the operating system, which has been a bunch of humans with antiquated tools trying to patch things with spreadsheets and clumsy collaboration.

Google provides a useful analogy for what Aera sets out to do:

When Google said way back when, ‘We’re going to build a system that helps people to find whatever they’re looking for on this thing called the World Wide Web, their premise was super-interesting. They said, “We’re going to build a hot copy of every single website in our cloud, regardless of their shape and form. Then we’re going to make sure that if there is an update, that we get that update’. That’s called crawling.

Then they said, ‘We’re going to process all that data. We're going to rank, we're going to index, we gonna, do all this great stuff, and then we're going to understand the best we can what the query is and then very quickly give some options. That’s called recommendations and over time these will improve based on how people click and their patterns [online].

Our premise is very similar. We are crawling enterprise systems…we crawl the largest instances of SAP and Oracle, their planning tools, external data sources. Basically we make a hot copy of customers transactional data and we keep it live. Once the data is in our system, we harmonize, we index, we do analytics at massive scale and we derive from all that a Cognitive Data Layer.

You have a question or you need a product or this customer or this plant, the system allows search. You want to do visualization, your control tower, your nice look-and-feel stuff, we can do that too. You want to do discovery in your data, we can do that as well.

Skills and knowledge

But the real differentiator comes from building skills on top of that data - and that’s not just a buzzword, says Laluyaux:

It’s actually a program that will replicate how people think about a certain issue. We’ve specialized right now in supply chain and manufacturing. When you think about a planner, someone whose job is to go and say, ‘I need to make sure that there is the right amount of product on the shelves in the store’ or ‘I need to build a promotion to increase the revenue for this product. How do I do it?’ or ‘I need to make sure that all my customer orders get fulfilled’ or ‘How do I plan what’s happening in factories?’.

Those people, they go in front of a screen and say, ‘What's up this morning? Oh look, I'll put through orders with no matching inventory. All right, so here starts my day - I'm going to prioritize the problems. Here is a big problem, so what do I do? Well, I need to find and follow a process. I need to go into the system and find if there is excessive inventory somewhere. For that I need a projection of my demand and my supply.'.

This is where the Aera system kicks in, explains Laluyaux:

You have an in-box called the Cognitive Data Workbench and it’s full of recommendations The system says, ‘I recommend you move inventory from this place to that place in order to close open orders. You have three days to do it and it’s going to save you $10,000. Would you like to do it?’.

We built all that process. We sit down with people and say, ‘OK, how do you run this business?’. And we run it in real time. There is a component of RPA [Robotic Process Automation] in what we do - even though we don't ever use that expression - but we've got to automate a lot of the thought process.

There are clear operational efficiency gains to be made, he adds:

If I tell you that to crawl the entire European instance of one of the largest consumer packaged goods company in the world, which has standardized on a handful of ERP instances with one vendor, to crawl and process three years of historical data takes 45 hours [with Aera] the first time and then it's immediate. If you talk to the CIO of that company, she said it would have taken 5 years to do that.

Away from the tech, macro-economic trends come into play here to make this a business-critical issue:

You look at all the tools sitting on the transactional systems and the knowledge that is required to operate those tools sits in people’s minds. If you're a planner for a big CPG company, you have learned over the years that when the temperature goes down by 10 degrees in a district, you need to increase the amount of Product X on the shelves because people consume more.

An example that was given to me the other day is that when the price of gasoline goes up in some districts, people will change their consumption of shampoo and conditioner. They don’t have extendable wallets, so they will use less conditioner and will shift shampoo brand. That kind of knowledge sits in the person beyond the computer; it’s not in the computer. Now you can apply all the data science that you want to come and derive that, but there is a lot of that knowledge.

This whole last 30 years, 40 years were predicated on the fact that people will be staying their job for a long time. But now you’ve got a new workforce coming in and after two years they’re saying, ‘OK, I get it. Where am I going next?’. What we are doing with this technology is capturing all that knowledge.

As for what’s next for Aera, Laluyaux makes the case that he’s taking a steady approach, but one based on the kind of innovation culture that comes from consumer-facing businesses:

Our industry, the enterprise software industry, is very, very old. We're getting excited by innovation that's not really smart…after that came the Google of the world and you look at their level of innovation and it’s unbelievable. They built a culture of innovation and it’s not complacent. These guys are real techies. We’re trying to build that culture.

The pioneers at SAP and Oracle, they really changed the world. They enabled globalization. But my initial premise for Aera…I was thinking that the next big challenge for companies is to reduce the number of layers. You think about the syndrome that we have of very large companies that have 9, 10, 11 layers of hierarchy. Why? Because we have no single visibility of the impact of decisions that we need to make. People escalate up, hoping that someone will be able to make the call. That’s the problem we can fix and it’s going to fundamentally change the way companies are organized.

My take

We’re keeping a close eye on Aera and its potential. The kind of ‘revolution’ that Laluyaux articulates isn’t one that’s going to happen overnight and has specific target markets, but it's coming, because it has to. He also makes a very strong point about the overuse of the term ‘innovation’ as applied to demos at trade shows that actually aren’t that innovative at all. That’s an idea I can certainly get behind.