I recently spent two days at the Ritz Carleton in Boston to get an update on the vision and product from the senior team at OpenText. I was not familiar with how far their world of Enterprise Information Management (EIM) had come in the last decade. For those not familiar with EIM there is a multi-billion dollar market for companies that help other companies manage their documents. In fairness, EIM has grown far beyond just managing documents and has become more of the critical lifeblood of a company’s command and control. Those ‘documents’ now execute and define the key processes that keep a company running.

For instance, consider the expense report that you will invariably submit sometime this month. It comes with a long list of rules for how it should be filled out and how it should be processed. Correctly processing that expense report is key (a small key) to the effectiveness of the company and also to you personally (so that you receive your reimbursement check!). Now consider that there are many such ‘documents’ and forms within your organization and between organizations (invoices, purchase orders, tax reporting forms etc.). They all need to be managed. Many of them are not well-structured data and are hard to manage with typical database systems.

For some industries and projects, the number of documents can be enormous. For one of OpenText’s customers that runs railways, it is estimated that there are over 5,000 documents that must be created, filed, validated and fulfilled for every 1 kilometer of track that is laid down. Tracking (pardon the pun) all of these documents is enormously important for the functioning of the company.

Because of this critical and growing role that EIM holds, OpenText CEO Mark Barrenechea made the bold statement that EIM would be as big or bigger than ERP in the not too distant future. This may well be possible and due, in large part, to the breakthroughs in Artificial Intelligence (AI).

Example: AI in the Legal Industry

A great example of using AI in a document/text based industry is in doing legal discovery. 70% of the cost of US litigation is driven by discovery and 70% of that cost goes to paying attorneys to review documents (approximately $1 / document).  Interestingly, the case never goes to trial most of the time. 98% of all US Federal Lawsuits are settled out of court. These settlements are made on the basis of the results of the discovery. So the legal discovery process really becomes the ‘trial’ in almost all cases today and sifting through documents to find nuggets of supporting information is the main engine that wins or loses most cases of litigation.

But this massively important task, that takes up massive amounts of very expensive lawyers’ time, is ripe for disruption with AI.  Machine learning algorithms used in AI have become fairly adept at reading and understanding documents. Though not perfect they do have the valuable ability to be aware of their own ignorance. That is to say that the machine learning algorithms can flag documents with either high certainty or with significant doubt. This can be a tremendous time saver for human lawyers. They can accept the AI’s decision when it is certain and pass off the document to a human expert when the AI is not 100% sure.

In one example, lawyers had to review only 2.9% of the total number of possible pages during a discovery process (which ended up saving 93% of the discovery costs). They started with 48 million pages that were possibly relevant to the litigation. These were then culled down to 12 million through some simple rules. An AI was then applied to further reduce the number of possibly relevant documents to a half a million pages. AI was then used to automatically redact the documents down to 141,000 possibly relevant pages that attorneys had to look at. Eventually only 142 key pages were used in the trial.

Would you like to speak to an Eckerson Group expert about AI?

Why AI is Such a Good Match for EIM

OpenText and other EIM vendors have a secret advantage over vendors in other industries. They deliver product within what I call the ‘path of action’. The path of action is any process in a business that absolutely must go on right now otherwise the business begins to fail. A critical web server would for instance be in the ‘path of action’ since if it failed the website would fail. On the other hand, a server supporting a data warehouse could fail without much immediate harm except that some reports might be late.

EIM is often delivered within the path of action. It manages, distributes and validates the documents that capture the real time processes that run the business. Some of these paths are highly manual and document processing and natural language text processing have recently been a hotbed for breakthroughs in AI. This triple threat (path of action, highly manual, AI breakthroughs) combines to provide a uniquely great place to successfully apply AI. And because it is in the path of action it will become mission critical and fairly and aggressively measured (which is a good thing because management will be paying attention).

OpenText is delivering AI in support of EIM to do things like smart document search and even automatically redacting of textual content to preserve privacy. Their CEO Mark Barrenechea has claimed that EIM will be as big as ERP is today.  A big ask but not out of the realm of possibility. My guess is that the distinction between EIM and ERP will begin to blur as AI allows unstructured data (especially documents) to be utilized to run the business just the way that ERP has utilized structured data over the last three decades. More than likely though, the term ERP will absorb EIM just as it has absorbed MRP (Materials Requirement Planning), SCM (Supply-Chain Management) and is now even consuming front-office systems like CRM (Customer Relationship Management).

Why the Cloud is Such a Good Match for EIM

In order to deliver AI as part of a critical enterprise-wide process such as enterprise information management requires a good solid and secure infrastructure. Since these documents are in some ways running your company, security is very very important and so is up time. The costs for downtime vary greatly but Gartner and Avaya have estimated the costs to be between $100,000 to $600,000 per hour. Some individual cases are much greater. One customer in the medical electronics space noted that they could lose up to $1 million for every minute their systems are down.

EIM vendors need to create secure cloud platforms or leverage existing best-in-class. OpenText is doing both. Utilizing Google Cloud and its best-in-class security systems as well as providing their own cloud resources.

OpenText has committed to four nines of reliability (99.99% uptime) and have strong partnerships with SAP and Google. Because of their acquisitions they have a very broad offering and can provide a ‘single SLA’ solution – or ‘one throat to choke’ accountability which is critical for large organizations. This is similar to Infor’s ‘industry in a box’ idea which I wrote about here: https://www.eckerson.com/articles/delivering-an-industry-in-a-box.

The Most Powerful AI is Deployed as Sprinkles

One last philosophical thought that was inspired by the conference. I’m from Boston and one of the local quirks up here is that we call the chocolate sprinkles you put on your ice cream ‘jimmies’. Supposedly this is because the name of the guy who originally ran the machine that made them back in the 1920s was Jimmy. Anyway, I’m originally from New Jersey so I’ll still call them sprinkles.

So what do sprinkles have to do with AI? Well, the extended metaphor here is that many implementations of AI are trying to come up with a new flavor of ice cream rather than just making your current scoop of mint chocolate chip a bit tastier. While self-driving cars and grand-master-defeating Go programs get all the press and are important (https://www.eckerson.com/articles/commodity-ai-and-the-next-best-experiment), it is the simpler applications of AI that are sprinkled on top of our everyday processes that are having the biggest effect.

This is why I think OpenText and the other EIM vendors have such a big opportunity and why Mark Barrenechea might just be right with his prediction that EIM may one day be bigger than ERP. AI is most powerful when it is sprinkled in on top of something that is already working and AI can just make it better.

When we look back at EIM ten years from now we may not see any big AI breakthroughs and we might scratch our heads as to why it has become so successful. But like the original “Jimmy”, tirelessly working at his machine to produce those sprinkles that make all ice cream more delicious, we may well see all of EIM (and probably ERP) steadily march along with incremental improvements all due to AI sprinkled in on top of what was already pretty good to start with. AI just made it a little bit better.

Related articles: 

• Four Signs You Need A Data Science Platform
• Don't Underestimate the Power of Stupid Artificial Intelligence Algorithms
• Commodity AI and the Next Best Experiment