Uncovering hidden knowledge with AI

Knowledge management has always been an essential aspect of any professional services organization—but until recently, it hasn’t been able to tap its full potential.

For years, knowledge management has been powered by enterprise search, which relies on explicit knowledge as its foundation. The problem is that every organization also has a wealth of implicit knowledge around not just the documents themselves, but how people interact with those documents and the relationships that exist around those documents.

Unfortunately, while search engines are very good at working with explicit knowledge, they’re not very good at spotting relationships—and not taking this implicit knowledge into account when conducting a search is akin to searching with one eye closed.

This is where AI comes in. In addition to using explicit knowledge and formal metadata, AI can leverage implicit knowledge to help professionals better uncover knowledge within the firm. For example, AI can look within a firm’s document management system to see what files people access or download most often, or it can look at the time and billing system to see which matters or clients a professional devotes most of their time towards.

Tapping into this data provides a series of “signals” that AI can use to better determine where knowledge and expertise can be found across the organization, assisting knowledge management in powerful ways and helping to uncover information that might otherwise remain hidden.

Find the Expert

Suppose you need to quickly find an expert on “European telecommunications” within your firm. Running a keyword search against CVs is a perfectly viable way to find that expert—if the CVs are actually up to date.

Likewise, sending a mass email to the firm asking if anyone has expertise in that area might turn up an expert—unless the person who actually is an expert doesn’t get the email because they’re out of the office, doesn’t respond to the email because they’re too busy, or simply doesn’t “raise their hand” because they are too busy focusing on their own billable work.

AI circumvents these roadblocks by tapping into dynamic data that is automatically generated, allowing it to draw conclusions about who’s an expert on what.

Let’s revisit the scenario of trying to find a European telecommunications expert. Instead of running a keyword search on potentially out-of-date CVs or emailing the entire firm, AI can analyze the time and billing system to see who has worked on multiple European telecommunications matters over the past two years—and maybe is even working on one right now.

A quick scan of the document management system will allow AI to identify someone who has authored multiple contracts related to European telecommunications—a strong signal that they have expertise in this area.

The beauty of this approach is that information in the systems is constantly being updated. Drawing upon this dynamic data means that AI can constantly adjust its rankings to provide a thoroughly up-to-date suggestion on who is an expert on a given topic.

Better yet, AI can provide insight into how it reached its decision rather than remaining a mysterious “black box.” If the AI says that Person X is an expert on European telecommunications, it can show the various factors it took into account and how it weighted those factors—giving professionals the benefit of understanding how the decision was made.

Since there is no “one size fits all” when it comes to determining an expert—different companies will naturally assign importance to different factors – the best AI solutions allow users to adjust how factors are weighted and to fine-tune the algorithm so that it delivers the most relevant results for their organization.

It’s important to note here that AI allows users to find experts not just more efficiently, but without compromising security. For example, when looking for an expert in European telecommunications, there might be documents that specifically reference this subject but are locked down with need-to-know access because the matter is particularly sensitive. AI can scan the documents and use the knowledge about its contents to conclude who is an expert in that area, and then provide the contact details of that person—all without exposing any of the sensitive content itself. For the firm, this means experts can be identified both speedily and securely.

What’s the Best Template?

Much like finding the best expert, finding the best possible example of a particular type of document can be challenging. If you’re putting together a loan agreement, an employment contract, or a real estate transfer agreement, which of the hundreds—or thousands—of examples within your organization should you use as a template? Which one is the best?

Sometimes, a professional can recall something they worked on in the past few weeks that would be perfect to use as a starting point. More often than not, though, users don’t have any specific document in mind.

Once again, AI is available to lend a helping hand. Algorithms interprets various signals around document activity to help determine the best example of a particular document to use. At one firm, the “best” template might be the one that was downloaded and used the most; at another firm, it might be the template that has been used most frequently within the last three months; at still another firm, it might be the template that has most often been selected by senior partners rather than junior associates. The point here is that the algorithm can be tuned to use criteria that the organization deems most important.

Alternately, professionals can use AI to sort thousands of documents into meaningful clusters to help zero in on the perfect template.

For example, say someone at a firm is looking for the “best” employment contract to use for a Latin American M&A event. AI can look at thousands of employment contracts, examine the similarities and differences, and group them into useful clusters. From there, the professional can look at cluster A, see if it’s a good fit for the matter at hand, and if not, move on to the next cluster without having to review the rest of the documents in the first pile. The result is a more streamlined approach to finding the perfect document—and less time “fumbling around in the dark,” hoping to find an ideal template.

After finding the best document, AI can go one step further and automatically “break” that document into components and pull out certain key clauses. In this way, it’s easy for a firm to build a clause library around, say, Latin American mergers and acquisitions in much less time than it would take a professional to manually find and pull the information—a boon for professionals who can now spend their time on less repetitive tasks.

The Power of Personalization

Personalized search results provide the information that is most relevant to the person doing the search, and that’s a good thing. After all, someone who works in the employment practice group doesn’t necessarily want to see documents from the construction practice group when they conduct a search—they want to see results that are most relevant to their daily work.

As users interact with the document management system, the time and billing system, and other enterprise systems, AI can better understand which clients and matters are most relevant to a professional—and even which files and documents—by paying attention to the various signals that are given off. These signals include what topics a user has searched for in the past, what files they’ve clicked on or downloaded, and so on.

Using these signals, AI can automatically tune its relevancy algorithm to provide richer, deeper, and more precise results to the user—and it will continue adjusting and refining its algorithm over time, as it is fed more signals.

Critically, having access to personalized results does not mean losing access to the entire result set within the organization. There are times when someone in the employment practice group might actually want to search on items from the construction practice group, for example, or from other practice groups—and that ability remains intact. AI-powered personalization simply reorders the document set to present the ones that are most relevant to a professional first, putting the information they need where it benefits them the most.

A Smarter Way to Work

AI’s ability to harness implicit knowledge is transformative for knowledge management, helping professionals see a fuller, more detailed picture of their organization than what explicit knowledge alone can deliver.

Using AI to quickly uncover hidden knowledge and expertise throughout the organization does more than allow the firm itself to operate more productively and efficiently—it also enables firms to deliver better service to their customers by drawing upon a richer, deeper pool of knowledge. Whether finding the best expert, the best template, or the most relevant search results, AI helps firms take their knowledge management efforts to the next level.