Spend Matters welcomes this guest post from Julien Nadaud, SVP of Innovation at Corcentric.
There have been significant shifts in computing every 10 to 15 years, starting in the 1980s with personal computers, then the internet in the ’90s and more recently the advent of mobile in the mid-2000s. What’s next for computers and the digitalization of the world? After the internet, artificial intelligence (AI) looms large.
The internet allows us to store and share information and knowledge in real time, accelerating the globalization of economies. This has profoundly changed the way business, in general, and procurement, in particular, are done. Now mobile devices allow us to have access to everything at any time, wherever we are.
In the early 2000’s, “procurement” became “e-procurement,” and most business-to-business transactions started to shift from paper, fax or modem-related EDI to online collaboration, negotiation, transactions and now payment. This is where most e-procurement providers were born, taking advantage of the combination of web applications and internet networks.
The gold rush started with B2B marketplaces, which were supposed to centralize all transactions online (and make a lot of money). Then the industry shifted its focus to digitizing and streamlining procurement processes inside the enterprise, as well as performing online e-auctions.
Now we are able to combine vendor management, strategic sourcing, contract management, e-procurement and AP automation with supply chain finance and payments on one unified business platform. The next step is to completely remove the friction of B2B transactions by combining the buy side with the sell side. In other words, managing full order-to-cash and many-to-many relationships using a business network.
During the 2010s, the pace of change accelerated exponentially thanks to new cloud computing, enabling business solutions to become fully part of the internet. Only providers that have invested in cloud and SaaS solutions can succeed in delivering innovation at the pace demanded by the combination of technology, globalization and societal changes.
Of those, only the ones with powerful and mature cloud platforms-as-a-service can offer the capabilities to manage complex business processes between many entities — buyers and sellers. At the same time, there is an increasing number of specialized niche solutions dedicated to either specific industries, regions or services. Bringing together those two types of entities is what the market is looking for now.
While all of this is going on — keeping everybody very busy with digital transformation and process automation (and playing catch-up most of the time) — another revolution is gathering steam: artificial intelligence, or AI.
AI is accelerating transformation.
It’s important to start by making sure we don’t misunderstand what artificial intelligence is. Humans like to compare themselves with AI and keep stressing that AI will never be our equivalent. This completely misses the point. AI today is just a new generation of computing that happens to be very good at handling any kind of information and making predictions.
If we compare humans to standard computers, there’s no argument that a computer doing billions of complex calculations every second is way more efficient than a person trying to do the same thing. How long would it take for you to divide two big floating numbers? But, that doesn’t mean that the computer is smarter than us.
That being said, let’s find out where AI shines, and why it is going to transform our industry.
First, AI can understand our language.
This is called natural language processing, or NLP. This field has experienced major breakthroughs in the last two years, giving computers the ability to intake any document in any language and understand the meaning of it. What’s even more impressive is that computers can now generate documents in natural languages, and we’re also starting to see images, music and videos being produced by AI.
Now we need to put that in perspective. Between 80% and 90% of the information in a given organization is unstructured — contracts, procedures and notes, marketing materials, etc., in many languages, including pictures, diagrams and videos. All this is the domain of humans. These items have been generated by, and are used by, humans, even though most of them are now digitized and stored on computers.
The fact that artificial intelligence can now access these items means that AI can learn from them and use them, and even start updating or upgrading them, and soon generate new ones in the future. This completely changes what computing has been used for in the business world.
Second, AI has access to a lot of data.
The internet has been used for more than 20 years to share information. Now, millions of terabytes of information are available online, and the amount of information is growing at an exponential rate. Big data technology, which has been developed relatively recently, can process very large amounts of data in near real time. The best example of how this information can be used is Wikipedia. It is probably now the main source of knowledge for new generations of kids, but also the main source of information for AI. Being able to read entire Wikipedia webpages in all languages enables companies like Google to develop pretrained models that can understand our different languages.
In fact, these models, called transformers, convert all human languages into one new digitized universal language that can generalize how humans describe the world. If you ask your Google assistant any question, it will answer based on what it learned from Wikipedia and other online sources of information. These transformers are open-source technology, and can be reused to develop very specialized and useful applications.
If you translate that to our business world, think of what decisions and actions you have to make every day. Each is based on a very specific context, and you have to rely on the information you have along with your experience. AI is very good at understanding very specific contextual situations and identifying the related information relevant to making the right decision.
This information may come from the past events or what has been configured into your business systems, but it could also come from the news, market research, business databases and online catalogs, or even discussions, meetings, emails or notes. And in this regard, computers have access to way more data than we do.
Third, AI is very good at finding patterns and making predictions by looking at all this data.
We humans like to think that we are unique and make decisions on our own, but in fact we make decisions based on what we know and what we have been taught to do. Computers can now detect these patterns and predict quite easily what will come next. It means that AI can be used to analyze our data and anticipate the outcome of it — this is called predictive analytics, and it’s becoming very popular in business intelligence tools.
But if you take it a step further, this technology can also be used to help us make the right decisions. If, by looking at all our options, the system can tell us what the impact will be of each option on the business, then it will recommend the best decision based on predefined business goals. This is called prescriptive analytics.
And you can guess where I am going: If the computer can tell you what to do, why not let the computer make the decision?
Last but not least, by definition, “machine learning” learns and evolves, which is the main difference between AI and the type of software development we are used to.
When you implement software, you have to wait for updates to get improvements. When implementing AI, the more you use the system the better it becomes without having to wait for developers to add new lines of codes. SaaS applications allowed us to provide quick updates to our applications, but machine learning is now a new paradigm where the speed of change and adaptation is brought to a completely new level.
The other disruption is that machine learning can learn from data that may be proprietary to a specific organization. Data privacy and confidentiality is critical in our business, but thanks to the same logic as the transformers used in natural language processing, we can translate information from different organizations into patterns that are aggregated into machine learning models that do not contain any more identifiable information. This is called generalization.
Using data from many sources, these models can create very powerful and efficient value. For example, when we train a model to read invoices using millions of documents, it learns how to read an invoice like a human. It does not store information about each invoice, but at the end, the model can be used to process invoices in a completely secure and confidential method.
AI has already changed our industry and how we do business.
Artificial intelligence is impacting all facets of business, and this change will come faster than anything we have seen before. It will give companies that embrace it the ability to move faster and be more competitive than others.
The good news is that AI will be easy and transparent for employees, because it is best at understanding and mimicking the way we do things, and helping us in our day-to-day work. It will free us from tactical and recurring tasks so we can focus more on strategic opportunities to improve the business.
In fact, the overall increase in efficiency should create more business, which means more jobs and opportunities for those willing to fully incorporate artificial intelligence.
Website of source