Optimize Operations, Increase Enterprise Value, Disrupt Your Competitors
Becoming AI-Driven is the key to growth and competitiveness
Photo by Jussara Romão on Unsplash
Introducing the AI-Driven Blueprint
Ian S. Wilson
Table of Contents
AI-Driven companies out-perform non-AI companies
Every industry is under attack by AI-Driven competitors
This a large US hotel-chain. You'll see these examples in every industry across different businesses and what we're showing here, kindly produced by CB Insights is that one major enterprise has lots of different parts to it, lots of different services and products, but for each of those services and products, there's a startup that delivers that service and often better than they do. So while none of these startups are doing the entire service offering of the enterprise, they're all biting away at it.
Here we see a major card issuer, a large US bank. And for each of the services it offers, Amazon, in this case, is offering a very similar service and we see it with other tech giants offering and expanding their own portfolios of services.
AI-Driven companies are valued higher than non-AI companies
Big tech companies (most of which contain a LOT of AI now) are generally valued much higher than non-tech companies.
Market value is now more intangible than tangible assets
Also, if we look at the components of the S&P 500 by market value and compare 1975 to 2020, in 1975, only 17% of the value of those companies was intangible assets. The rest of it was tangible assets like buildings, plant machinery, stocks, resources. In 2020, 90% of the valuation of those top companies was intangible, which, beyond things like IP and brands, was data, analytics and other software. But the key point is that an enormous percentage of the valuation of businesses today are in those assets.
It only took six years for Ant Group to become the largest bank in the world
Here's a look at Ant Group, a principally AI driven company. I looked at this chart a couple of years ago, and they were already valued as the 10th largest bank in the world. If their IPO had gone forward last year, they would have been valued at as the largest bank in the world. Why is that important? First, they're 100% AI driven and second, they're only six years old! Six years old to be the largest bank by market capitalization in the world is incredible .
AI-driven companies are massively efficient
Now look at the size of their businesses in terms of employees. Compare Visa, Mastercard, PayPal, and Ant Group to the Big Four American banks. Citibank alone has over 300,000. Tech and AI-Centric firms are driving far more revenue per employee than non-tech AI-optional firms.
What does it mean to become AI-Driven?
Fear of the unknown is preventing massive action
Many businesses know that they want to become more AI-capable but they’re not moving? Why is that? Because in front of them there is a lot of uncertainty - the fog of war. It may be nice to aspire to be more like the FAANGs but how?
But what's the path in between? How do you get there? All of the unknowns between where you are and where you are trying to take the business are uncertainties and those uncertainties create business risk.
What are the 16 UNCERTAINTIES that are creating RISK?
How do you reduce those risks?
We have a risk reduction framework to help businesses understand and mitigate the risks of becoming AI-Driven Enterprises.
The four pillars of risk reduction
There are four key pillars or principles of risk reduction
Why bother addressing the risks? It will build support and potentially investment from our enterprise leadership teams. Many initiatives die because they never have enough executive support. Why? Because there's too much risk. If we mitigate the risks, we can increase the investments.
AI Vision: Where are we heading?
Doing lots of data science is not the same thing as being AI-Driven. It's a component of a much larger whole.
So what does it really mean to be AI-Driven? We believe that there are two key elements. On the one side, you have decision automation which is going to increase bottom line value, increase margins and revenues, and reduce costs. On the other side you have data and asset monetization. Most of the value of modern-day technology companies is intangible and much of it is derived from data and analytic assets. This increases enterprise value.
So decision automation improves the income statement and data asset monetization improves the balance sheet.
Decisions are a fundamental input of a modern business
The core principle here or any business for any business process is you start off with a desired world. You write a business plan, you write a forecast or projection. How do you want your business to actually operate? What do you want it to do in the best case?
Then that meets the real world, the current world, and the problem with the current world is that it doesn't do what you expected! So you continuously have to look at how your business is performing in the real world and look at what can you change to either fix problems, optimize or take advantage of opportunities.
And you continuously go around that very simple circle. Look at our desired world. Look at what the real world is telling us are our sales down is our market smaller than we expected bigger than we expected. What change can we make to optimize this?
AI-Driven companies have digitized their decision making process
What does that look like in when you're AI-Driven? We essentially digitize that process. Our desired state really doesn't change fundamentally, except we will put those values in some digital storage somewhere. But key to this are the two components at the bottom, a digital twin and AI decisioning or decision automation. And the key with the digital twin is we need a representation of the world. So if you're an oil refinery, it's a digital twin of an oil refinery. If you're a retail store, it's a digital twin of your store and your customers and your market and your sales, only those components you care about. And that has been fed by digital data streams that are feeding that digital twin and continuously updating it. If you have all of this in place, then an AI decisioning system can look at that real world, compare it to your desired world and say, okay, what's the difference? Is it good? Is it bad? How can we change? How can we optimize? How can we fix problems? But the key here is that loop should be automated. And that's the end goal. So the key there is the decisioning system is able to automatically look at the differences and try and figure out what's the best route forward to make the real world look more like your desired world. This is a very simple core principle, but incredibly hard to execute.
Also, it may take some time to be fully automated in every part of your business, but certainly some parts of your business can start off being fully automated and you can gradually expand out.
Isn't RPA enough? In short, no. RPA is great at automating business processes but within those business processes lie tasks and within those tasks there are decisions that are beyond the ability of the RPA team to handle. It's not about RPA OR AI. It's about building RPA WITH AI.
Better, cheaper, faster, with more options
So at the end of the day, decision automation delivers decisions that are cheaper, faster and better but it also does this by reviewing an increased range of options. With a person you might only have time enough time in the day to look at 10 different options and only look at 10 minutes worth of detail for each option to make a decision.
A machine could look at 10,000 options and review every piece of detail in 10 seconds. So you can look as deep as you want as wide as you want. That's an incredible benefit!
Decisions at the speed of AI
What does that deliver us? It gives us decisions at the speed of AI. These are a thousand times faster, a hundred times cheaper, 10 times more effective, that are better quality.
What you need to think about is if your business were running and making decisions like that, a thousand times faster, a hundred times cheaper, 10 times more effective, what would your business look like? What would it look like today? What could you make it in the future?
Data & Analytic Asset Monetization
Companies know the value of their chairs, but not the value of their data
Today there is probably somebody in your organization who can tell you the exact value of all of your chairs. But there probably isn't somebody who can tell you with any reasonable certainty, the value of your data or your analytic assets. That's because we don't have the infrastructure in place or the governance or the mechanics to look at that. Now that is amazing because for those tech businesses, most of their value is based on that. And most companies today have no idea what that is.
The two fundamental building blocks of decisions must be digital and governed
What are these things? Data assets are basically codifying the knowledge in your business. Analytic assets codify the expertise in your business. These are the two fundamental components that allow us to make decisions.
If these assets are digital and governed, then they can be monetized, that can add significant enterprise and market value.
However, if those assets are analog or the ungoverned, which means you have no idea what data you've got, you have no idea what analytics you've got, they cannot be monetized and they're basically worthless. But more than that they're treated as a cost, so you're storing the data somewhere and paying for it and treating it as a cost to be managed down.
Finally, if that knowledge and expertise is only in someone's head, that person can leave the building and take that value with them.
It is critical to identify, codify, and digitize and govern these assets - they are the key to faster decision making!
AI should really be thought of as a set of tools, technologies, and processes that help you deliver on your corporate and business strategy. We break that down into three key elements below:
AI Capability Strategy
A core component of AI strategy is Capability Strategy. That is made up of made up of three key elements:
Maximal reuse of assets to maximize marginal ROI
10,000 use cases (or more) is too hard to manage
Capabilities serving use cases
Maximize re-use to maximize marginal ROI
It is absolutely critical that we maximize reuse of expertise, platforms, services, data, and analytics to maximize marginal ROI. Cost is the enemy of AI today. You've heard time and time again "Where's the ROI with AI?" The lack of ROI is because we have large fixed costs when we're introducing something. That's because we have to build the foundations, the platforms, the tools, the teams, etc.
Most businesses approach this in a way where the first use case out of the gate has to pay for that fixed cost and that is why they never achieve any real ROI. It's critical to spread those fixed costs across multiple use cases and teams. So if you introduce a new platform, build new datasets, cleanse it and prepare it and build analytics with it, you want to reuse those as much as possible and spread the cost. That's how you capitalize or amortize the development cost by spreading it across every use case possible.
In short, don't solve one use case (contract management for procurement), solve for an enterprise-wide capability (document processing) that can serve multiple use cases across the enterprise. That dramatically reduces the cost base and allows you to actually potentially achieve good ROI. This is fundamental and it's not happening often enough.
Use cases are really decision points and there are 10s of thousands of them
Let's have a quick word on use cases. It can be distracting and confusing to focus on use cases. There could be literally tens of thousands of use cases in your organization which is why it can be counter-productive and confusing to make them the center of your efforts.
Often the business use case is a whole customer journey, or it could be a business process or a task within that process. In the general case however, an AI use case is a single decision task within a process. So it looks like this:
Customer Journey contains...
Business processes which contain
Tasks which contain
Decisions (<-- this is the AI Use case!)
So what does that look like in a bigger sense? Imagine that we have a data stream coming into our business and it's email. And we're going to process and ingest all of that email, and we're going to look for customer complaints. So we build our AI platform. Now there might be only one group in the business that cares about that - the customer complaints department!
So we process all of that email, and if we find a complaint, we send that email to the customer complaint department. Now imagine in the diagram here below, that's one of those blue blocks. The problem is that customer complaint department has to pay for the whole thing. They have to pay for the project, the infrastructure, the platform, all of it. That means they really struggle to find ROI.
Capabilities - an alternative approach to delivery higher ROI
Instead of solving one use case for one department, we could build something much more horizontal that can look for anything we want to find in email right across the business for everybody at once. This could be customer complaints, sales orders, bad language, insider trading, purchase orders, contracts, you name it. Let's find and flag any kind of document in email that the business cares about and send it to the right deparment. In the diagram below that shows all of those lines going to all of the blue boxes or departments.
This brings us to program strategy. What we've built here is not a use case, it's an enterprise-wide AI capability to read email, understand what it means and forward it to the right people. That's a great capability. By spreading it across the enterprise to all stakeholders on our diagram below, we can ask them all for a bit of funding.
Now imagine we've got now a whole portfolio, perhaps, 10 or 20 different use cases for all of those people here then within all of those, typically your finance department will expect each of them to produce a little bit of benefit. However, what you'll find in general is that some of them will perform better than others. Some might perform outstandingly, some might fail completely. Some will be in the middle. That's actually the way a venture capitalist manages their portfolio of companies, and our little portfolio of use cases is just like that.
So what we're looking to do now is to balance the ROI across the portfolio of use cases (all derived from one capability), not just a single use case. So in the approach where you go use case by use case, you've got to get out of life for every single one. If you build a portfolio, you can balance the ROI across all of them, which is a much easier task.
To sum up, don't solve for a single use case, create enterprise wide capabilities that solve many use cases at once. Then crowd-source the investment funds from all those teams and balance the ROI across of them.
Finally, we get to delivery strategy. We've got our program but how do we deliver it? Or rather, how will the organization deliver on this?
To answer that, it's important to recognize that becoming AI-Driven is about making a full AI transformation. That is the third stage of a three part transformation that has been doing on for over a decade. It starts with digital transformation. Once you're digitally driven, meaning that the data in your business is digital, not analog, then you can start accelerating your data transformation. That means getting your data ready, because if you want to be AI driven, you need great data.If you haven't done an industrial transformation of your data, your AI transformation is really going to struggle. These things work together as a whole, they leverage and build on each other, they work better than the sum of the parts.
The key there in terms of delivery is that we'll use our transformation programs to do our delivery, not a specific individual AI group, because we already have our transformation groups! Let's make use of them.
Strategy can optimize, add value, and disrupt
Once we deploy this overall AI strategy, we can optimize, add value, and disrupt. Why does that matter?
Because operational efficiency empowers the COO. We want the COO onboard.
Data and analytic value empowers the CFO. The CFO is going to love that. So we want the CFO onboard.
The capability to disrupt empowers the CEO. And we definitely want the CEO.
So let's try and build something for each of those.
What we're trying to say here is you wouldn't try and do fly a plane without an expert. You wouldn't try and do open heart surgery either. Many organizations try and do AI without experts and it doesn't work out. It is critical to put in a Center of Excellence up front that delivers expertise.
The 4 components of an AI Organization
The mission of the CoE is to "Support the realization of benefit from strategic, industrial, enterprise AI" by enabling the organization to do the following:
Industrialize: It's looking to industrialize the AI process with best practice, and with training and expertise.
Accelerate: You can do that by building reusable assets. So everyone has a central place to go to find things like frameworks and operating models, training and tools. They don't have to figure it out themselves or reinvent the wheel on every project.
Scale: You want to scale the AI process with operating models, capability assessments, use case evaluation frameworks, delivery frameworks, and more. Then your team has everything they need to do their work.
Govern: We want to build and govern all of these initiatives across the group, so that they're not random and chaotic and we can make sure they're strategically aligned. They're reusing all the assets, and we are capitalizing the costs.
Most companies have 25 different AI groups doing 25 different projects on 25 different platforms, many of which are duplicated. We don't want that, what we are trying to do is realize the benefit from strategic industrial enterprise AI.
By building a Center of Excellence, you can industrialize, accelerate, scale, and govern your AI efforts properly.
How a COE works
A CoE operates right across the business. So it operates right from the board, through the business lines, through transformation and down into Analytics, Data, and IT.
It touches every area from platforms and architectures, to transformation journeys, to innovation, right through the board level to regulation, risk and audit. It is the glue that holds everything together.
Evolution of a CoE
We generally start with a centralized CoE model, where you've got one group that's controlling all the initiatives around the business.
Once the organization is scaling up efforts, they often to move to a hybrid CoE where you've got a central group that's perhaps monitoring governance and training, but then you've got smaller CoEs around the business that are helping delivery teams, various models, depending on the stage of your business.
Eventually some companies move to a fully federated or distributed model but that's when they're very mature.
CoEs drive collaboration
One of the key aspects of a CoE is collaboration and what this really means is that it's here to help enable collaboration between these different stakeholders, rather than them all going off in separate directions. The CoE brings them together and leverages them all and brings that collaboration. So whether it's platforms, transformation, data, or analytics, or IT, the CoE should be uniting them all.
Note: This is why it does not make sense to have a dedicated AI delivery team in any business. You already have existing teams with lots of expertise and they should be leveraged. Don't reinvent the wheel. AI is hard enough without that.
You need a center of expertise and it should be delivering expertise, NOT PROJECTS. You have delivery teams to do that. And if you have a CoE, that's delivering projects, it will never scale and will hamper your long-term efforts.
"Let's just get some quick wins on the board" is a common phrase in the boardroom. It assumes you haven't already found them but your team probably HAS found them and won them already. The bigger problem is that, at this stage of the game, this is just the wrong mindset. It hopes for quick, cheap, and easy and there almost isn't anything like that. You need to have the right mindset that this is long-term, strategic and foundational to the future survival and competitiveness of your business.
AI depends on digital, data, and analytics foundations underneath
Your digital transformation delivered digital data. Your data transformation delivered data infrastructure, competency, and teams. Your analytics teams built their foundation on top of this and they are the ideal team to extend their skill base because up into the AI realm. And finally you have your AI CoE at the top managing the integration of all of the foundations below.
We want to look at digital in, digital through and digital out. What that means is we need all the data coming into our business and all the data going through and out of our business to be in a digital format, because we need it in that format in order to manipulate it. This is how we have the data to create our digital twin of our business. We need that digital twin to be as accurate a representation of our world as possible. So we want to feed in tons of data to be able to represent that.
Digitization of our business is the base.
Data is often described as the new oil. It is a really good analogy because it needs to be discovered, and extracted just like oil. It needs to be piped to refineries. It needs to be processed and stored before being piped out to business consumers - in our case analytics and AI systems. When we think of data, we've got an awful lot of infrastructure and architecture under the hood, but fundamentally, we've got just three key components we care about.
We have data producers, data processing, and data consumers. Somebody is producing the data, whether it's a customer or a market, a data feed, a satellite, wherever it's coming from. That data has to be processed. It has to be quality checked. Someone has to make sure it is fit for purpose because then it can be used by a consumer.
And that could be your analytics team, an AI system, or a business user.
Our analytics foundation is essentially the tools made available to our analytics people to take that data, to take that knowledge, and to build expertise around it. That could be an analytic model or a machine learning model or a set of business rules or a mixture of all of the above that delivers answers to business users.
So we need to do the modeling, and we need to be able to deploy those models into our systems so they can run in real time. Especially when we talk about AI, we're not talking about dashboards or printing reports, we are talking about real time systems running autonomously in the best case.
So our analytics people need those foundations, those tools and capabilities to build and deploy with.
Credit: Snowflake and Dataiku
So we now have a CoE, and an overarching strategy and at this point you can also begin to evaluate and select the tools and platforms that you will need to build, scale, accelerate, govern, and industrialize your AI efforts.
Tools and Platforms
AI is exploding and so is the global market for AI startups. NVIDIA just released their report of 8500 VC-funded AI startups with $60B of funding. FirstMark Venture Capital release v1.0 of their Data & AI landscape map in Sept 2020 and it's hard to make sense of it. The number of companies and open source projects are too innumerable to count. Who do you work with? Who will still be around in a year? How about three years? Which ones do you select?
HOW WE CAN HELP YOU BECOME AI-DRIVEN
Join our Masterclass in Business AI
The Business of AI
Setting the stage for the discussion ahead
How business can be transformed with AI
What are the business benefits of this transformation?
The Economics of AI
How AI changes business economics
The Macro economic affects of AI
The Micro economic affects of AI
The Politics of AI
The elements of AI driving politics
The Geopolitics of AI power
The Regulation and Ethics of AI
How AI supports and benefits Corporate Strategy
How AI supports and benefits Business Line Strategy
How AI supports and benefits Business Function Strategy
A framework for AI Governance
A framework for AI Risk
A framework for AI Compliance
Where AI structures sit in an organization
The AI CoE expectations, vision & strategy
The AI CoE structure, op model & delivery
What are the drivers of Digital Transformation?
What are the benefits of Digital Transformation?
How do we integrate Digital, Data & AI?
How do we become Data Driven?
What are the foundations of Data Transformation?
What is the strategy for our Data Transformation?
Evaluation automation fundamentals
Decision automation fundamentals
Action automation fundamentals
The current state of enterprise AI
Single industry focused state of AI
Current AI state facts and figures
The Technology foundations for AI.
The Data & People foundations for AI.
The Process foundations for AI.
How we define AI Capabilities.
A framework for AI Capabilities.
A discussion of example AI Capabilities.
AI Use Cases
How to evaluate AI Use Cases.
How to score each AI Use Case.
How to Rank a set of AI Use Cases.
The challenges of building an AI Portfolio.
How to build and get funding for a Portfolio.
How to deliver your AI Portfolio.
The AI-Driven Enterprise
The Design Principles in the "Enterprise OS".
The EOS Framework & Operating Model.
How to design an EOS core tactic / skill.
Get support and mentoring
Individual High-Impact Mentoring
Invite us in to work with your team
Lunch and Learn
Level-set your entire organization on how AI can support your business.
Build AI fluency and get team alignment on overall strategy
AI Program Development
Build your AI program and Center of Excellence in 90 days
Frequently Asked Questions
Q: Who is this for?
The AI-Driven Blueprint is aimed at Boards & Executives, Program & Transformation Teams, and Managers of all levels working on transforming their organizations into AI-First companies.
Q: Who is the practice area leader and why should I listen to him?
Ian Wilson is the former Global Head of Artificial Intelligence at HSBC. He has spent 25+ years successfully delivering business value with a diverse range of AI capabilities across a broad range of industries from Financial Services to Defense to Entertainment. Strategic Advisory & Consulting. This program is the culmination of his 25 years of experience building AI companies, products, and programs for the world's largest organizations.
Q: What’s included?
Depending on the engagement, we have a complete course curriculum available, as well as the option to have group and 1:1 mentoring and we can also facilitate workshops for executives, management teams, and entire companies.
Q: What if we're already started?
Most companies are already somewhere on the journey but we have found that there are often gaps and those gaps are reducing ROI and slowing down the organizations' ability to go beyond hand-built machine learning models and PoCs and "quick wins" to get to industrialized AI as a core competency. We will help find and close those gaps to help you accelerate your progress.
Q: How is this different than other "executive AI courses" or programs?
This is a full transformational blueprint that covers executive team alignment, initial strategy, team building, use case evaluation, program design and funding, center of excellence stand-up, and more. It can be delivered stand-alone in a course, with group mentoring, 1:1 mentoring, executive workshops, and company-wide engagement.
Q: How is it delivered?
We work across learning management systems, video conferencing, team-wide collaboration environments, and more. Most of our engagements are with distributed teams so we mostly work remotely. On-sites can be arranged as needed in particular situations.
Q: How much does it cost to work with your team?
Each customer and therefore each engagement is unique. Please book a meeting above and we will be happy to meet with your team to find the right mix of products, mentor groups, and workshops.