A few years ago I saw this headline news flashing all over the internet.
“Our dealers are missing up to $18 billion in easy sales.”
The Chairman and CEO of Caterpillar suggested that the company and its dealers were losing $9 – 18 billion in easy sales revenue as their sales, both internal and dealer networks, weren’t monetizing the real value of data.
He worried that..
“They are not tapping into the wealth of real-time customer data now at their fingertips; they are not communicating with each other, and they are not providing customers across the globe with a consistent experience when it comes to everything from e-commerce to parts and services pricing.”
Long story short, the whole idea was to convert the company’s mentality from dumb iron sales to data-driven, machine learning-driven sales.
So I read the press, thoroughly understood their strategy from their annual reports, read their investor decks, and then eventually wrote to the CEO.
But first I organized his dilemma and linked it to the value loss he had alluded to.
The company had been scrambling for additional revenue to circumvent a volatile economy and demand. Let’s say, I had a bit of an understanding of the global steel market from my first startup when we helped the world’s largest steel company with revenues of ~$120 billion so we knew where the markets were heading. It was no secret!
After peaking at $65.9 billion in 2012, sales plunged nearly 16% a year later as capital investment by the global mining industry tanked as expected. More sales warnings were on the horizon.
I thought it was time to map their defined global driver, which I’m sure a strategy consultant might have created for them, to data-driven goals and actions. I won’t share that here but you can see the rationale in the questions I tied up to these business drivers and eventually in my letter to the then CEO, Doug Oberhelman, I suggested the following three actions:
- Include the overall ecosystem and partner network in your data-driven, AI transformation strategy. In fact, include them all and try to capture both machine-chatter and consumer-chatter in a unified console.
- If you don’t have an AI transformation program underway, involve all those 178 businesses from the start. Ask your CIO to get you a current-state analysis of where you are!
- Educate your partner ecosystem so they can effectively use the digital & AI technologies and platforms to make more calls and secure targets within their territories. Just knowing about those IoT-ready billion machine parts is not enough, dealers and consumers want to know only what applies to their domains and territories. In other words, contextualize that chatter!
I think since then a lot has happened within Caterpillar and I’m hoping that they have made some progress by now. My intention was to scratch the surface and see if there was a way to start mapping and identifying potential data-driven MVP (minimum viable AI projects) where the machine and deep learning could be potentially applied to create services, solutions that could help both bridges the gap as well as build new bridges to the additional revenue stream that the CEO was hoping for.
So there is more than a dime to make out of data, that’s for sure!
DATA-DRIVEN JOURNEY CAN BE TOP-DOWN OR BOTTOMS-UP – YOU CHOOSE!
Need for a Chief AI Officer? Not really, you just need a competent executive who is business savvy and technically sophisticated about the machine and/or deep learning technology!
Many organizations do the top-down strategy where they hire strategy consultants. This involves a thorough temperature check into various aspects of their organization such as finance, marketing, sales, procurement, legal, IT, M&A, R&D, etc., and return back with multi-dimensional state analysis.
Top down-strategy essentially reports how far separated the firm is from being a data-driven AI company.
The top-down strategy is then translated to various domains, budgets are handed out and this trickles down to a multi-year program that eventually morphs into something great – if the firm has dedicated focus and is resilient to executive succession, or goes into Project Death Valley if folks start running around without goals, milestones and talent.
Bottom’s up journey is driven by domain experts in marketing and sales or machine learning divisions who see value in AI and go with it!
I have seen this happen as well where a simple data tech problem led to a global firm establishing a full-scale AI roadmap development. This firm actually went ahead setting up a huge fund to do a full-scale build (re-engineer their organization, processes, and tech) and buy (buy out players who had robust data and experiential products) play into the market.
One such CxO I worked for had also gone back to learning machine learning and eventually ended up addressing his whole global organization to adopt machine learning. This was a unique but very welcome approach as this drove excitement and engineers felt that they could change the world.
Great leadership who understands AI holistically can really turn around its organization dramatically.
IN THE AI ECONOMY, IT’S A WINNER TAKE-ALL GAME
It would be foolish to assume that you can hide under your pillow and not head out to be driving this change for your organization.
It might seem easy to sit and start analyzing firms at a distance but the problem is that most companies are facing massive challenges and the adoption of disruptive technologies in machine learning and deep learning can put them at the top of the food chain.
There is no way around it this time.
Most All companies and businesses are in the battle zone, and merely resting on new shiny platforms is not enough – push harder to move to disruptive AI platforms and adopt disruptive processes.
We at deepkapha.ai – (disclosure: my company) – take this huge disruption very seriously and ensure that enterprises, the startup ecosystem, and data scientists find and solve the real-world problems we have entered a collaboration with AnalyticsVidhya.com. Both Kunal, CEO at AV, and I believe that it is through solving real-world problems that true transformation will be achieved.
So you better start preparing to climb the AI Hill!
DEEP LEARNING HILL CLIMBING – DO IT TOGETHER WITH YOUR CO-WORKERS
Ok, enough lecture about what you need to do. This is part of my lecture/keynote tour across the U.S, South America, and Asia and I’d like to quickly introduce you to it.
Essentially every leader is asking if this thing called AI can be applied to their enterprise like it is at Netflix, Uber, Amazon, etc.
Notice, that is the top of this hill, we need to cover some ground before we get there. But fortunately, there are ways to get to the top of this. What are they?
Let’s find out!
PLATEAU 1 – THE FUNDAMENTALS
Brief Description: College level math linear algebra, probability, and programming are the fundamental skills you and your co-workers will need to get started with data science technologies and techniques such as machine learning and deep learning.
Benefits For The Enterprise And Employees: Enterprises need to pay good attention to understanding these skills for both internal recruiting and/or external hiring. Google, Facebook, and practically all other data-hungry firms know and interview employees with these talents. Having this fundamental understanding is key to develop intuitions where your co-workers will (hopefully) develop their own algorithms that can give your firm a huge competitive edge.
Ideal candidates/roles here have good potential to be algorithm engineers and statistical analysts.
PLATEAU 2 – DATA VISUALIZATION
Brief Description: Plotting tools, domain-specific libraries, and vendor and/or open-source tools to handle data.
Benefit for the Enterprise and Employees: Anything that we are toying with needs to be explained in as few words as possible so that our counterparts, who could be business partners, dealers, and suppliers, can understand it clearly. This is also an essential skill that involves not only understanding how data needs to be diced and spliced but also good presentation skills to try to related to your not-so-savvy peers.
Ideal candidates/roles here can be data analysts, visual artists or just any bored manager who knows how to do this but was previously never exposed to these tools 😉
PLATEAU 3 – MACHINE LEARNING
Brief Description: Various techniques such as sampling, clustering, regression, and classification have been used by your organization for decades. Only now these are getting to a point where you can scale since you have more and more data in your organization.
Benefit for the Enterprise and Employees: Machine Learning is a skill that seems hard to attain but if examined closely you will realize that you have been using these in your business for a long time. It’s just that it has been called something else, expert networks or simply an ETL task that involved a lot of preprocessing which your DBA would still know so well.
Ideal candidates/roles are your data engineers, statisticians, and I would even be bold to say your developers and architects. The last two are key in making machine learning systems being developed and served into your production environments.
PLATEAU 4 – DEEP LEARNING
Brief Description: When it is image, text, audio, or video data, you can employ various techniques such as computer vision technique called CNN (convolutional neural networks), text analytics NLP technique such as RNN (Recurrent Neural Networks), or other emergent forms of Neural Networks such as GANs (Generative Adversarial Networks) or the latest ones such as Capsule Networks.
Benefit for the Enterprise and Employees: Companies that are quick to identify and even observe their environment from the above tasks perspective such as vision and text, can already put their engineers to work. Face recognition, parking lot planning, floor space management, contextualizing a conversation, creating soothing music according to time of day or mood of your co-workers, solving pressing problems in medical diagnostics, traffic congestion problem solving, depression detection and solution, and so much more!
Ideal candidates/roles are biomedical engineer.
The wishlist for deep learning solutions is endless and is growing dramatically.
PLATEAU 5 – APPLYING IT TO YOUR INDUSTRY DOMAIN!
Brief Description: Applying all your learning from Plateau 1 all the way to Plateau 4 to your industry specific domain is where the rubber hits the road. And this illustration is just looking at how you can do a full stack data driven project in machine learning or deep learning.
Benefit for the Enterprise and Employees: I have left out a large chunk of the exercise about hypothesizing, conceptualizing, scaling, documentation, MVP (minimum viable project) identification, platformization, and eventually industrialization at scale across geographies and operating country business units. The benefit of doing this is a no-brainer.
Organizations that want to survive in this data-rich AI economy better pay attention and have an AIPlaybook or prepare to be decimated by competition.
IN THE END…
Businesses seem to be waking up to the call for applied AI but more needs to be done.
It still needs to be seen who is doing this as a window dressing exercise and who is serious in this game.
The steps to reach a mature level where organizations are fully equipped with the right talent to have an adequate and secure infrastructure where they have access to developing custom-build algorithms, and an executive board that is not only aware but actively supportive of this transformation seems daunting.
In my next article, I will talk in detail about how to identify MVP AI projects, run them and seek opportunities to find that unicorn project that you can industrialize at scale.
originally posted on Forbes.com by Tarry Singh