top of page

Why A.I.-enabled KM solutions are the foundation for law firms

“The most important contribution management needs to make in the 21st century is to increase the productivity of knowledge work and the knowledge worker.”Peter Drucker, Management Challenges for the 21st Century, 1999

Law firms are in the business of procuring knowledge workers (lawyers, paralegals, and professional support lawyers) and explicit forms of knowledge (for example: regulatory guidelines, legislation, templated agreements), and synthesising, value-adding to, and selling this knowledge in the form of legal services and other work products to clients [1]

In his seminal paper “The Nature of the Firm”, Nobel laureate Ronald Coase posits that firms will grow in size if they can lower procurement costs and internalize as much of their production as possible. Applying Coase’s theory to the practice of law, firms which can accrue large cost savings relating to knowledge aggregation, capture, and retrieval – collectively, knowledge management (“KM”) - will develop a competitive advantage.

Law firms seem to intuitively understand this, and have placed importance on collecting existing knowledge which can take the form of internally developed templates, closing sets for a completed transaction, legal opinions rendered in a previous matter. However, aggregation of knowledge is only the first step in an effective KM strategy. For KM to become a driver of a law firm’s profitability, it should aim to create an ecosystem of tools, processes and a culture that will efficiently deliver the right knowledge to the right lawyer at the right time.

Thus far, implementing effective KM has been considered a “nice-to-have”, or a luxury afforded only to the largest law firms. The main reasons for this are that the leadership teams of law firms have struggled to quantify the business impact; additionally, lawyers find KM tasks to be tedious and burdensome, even if they acknowledge the benefits.

We would argue that the time has come for all law firms, regardless of size, to re-evaluate their approach to KM: KM is no longer a luxury, but is an imperative for law firms to succeed in the current landscape. This change is driven by a confluence of factors: clients expecting to have legal services delivered efficiently, the advent of modern technology to handle Big Data, and the lowered cost of using these technological tools.

The Change: not all time is money; Time is revenue for law firms.

The basis of a law firm’s charging model is the billable hour. Even with more firms offering alternative fee arrangements, the billable hour still remains the reference point for how legal services are priced. With the billable hour model, there is little incentive for lawyers to spend time on thoughtful organization of work product from concluded matters. For example, in transactional teams where closing sets or transactions bibles have to be delivered to clients before the final bills are paid, it is typically the most junior member of the team who is responsible for creating this set, and often without consideration for electronic retrievability or search functionality thereafter. This is the case even though lawyers in transactional practices very frequently refer to closing sets as precedent documentation. In the absence (or inability) to retrieve relevant precedents, lawyers wind up spending hours trawling through poorly catalogued content and redrafting documents which may already exist. For as long as the client is happily paying for the time, the billable hour remains king.

However, clients’ expectations are rapidly changing. In a recent study of 1,400 legal matters across 400 law firms [2], General Counsels found that flat fees arrangements have been effective in reducing legal costs without compromising the quality of work. This led the authors of the study to conclude that the demand for flat fees would continue to grow. Through our conversations with law firms in London and Asia Pacific, we’ve seen that firms are experiencing this pressure in a slightly different manner. Lawyers say they are often pressured by clients to write-off activities which were previously billable. These include research (“Aren’t you supposed to know your work? Why are you billing me to ‘do research’?”), template selection (“You claimed to be experienced and yet you’re charging me so much to draft a simple contract?”), and project management (“Your firm ought to be using technology to improve efficiency!”).

If unbillable tasks are not managed more effectively, then the flat fee and other similar alternative fee arrangements will always be a squeeze on a law firm’s resources. It is here that KM initiatives can help attorneys be more efficient will be accretive to their firms’ profit margin.

In addition, KM can also help generate additional billable hours. The success of law firms is commonly measure by their “profit per equity partner” (PPEP), which is driven by averaged realized rate, margin, utilization rate, and leverage. KM can improve law firms’ leverage [3] by helping senior lawyers and support staff make the best use of their time, allowing them to deliver additional, high quality work [4].

The Challenge: no man is an island

While there is no agreement on the exact scope of KM, literature generally considers KM to span access the following activities:

  1. Capturing or documenting knowledge;

  2. Packaging knowledge for reuse;

  3. Providing access to knowledge via various retrieval mechanisms; and

  4. Utilizing knowledge to generate new insights.

The biggest challenge that we have heard consistently from clients is that the knowledge that they need to access are of different types which exists in silos and across multiple data stores. Examples of data stores include document management systems for documents and emails, matter management software for matter and billing information, content management sites for internal know-how, and external databases (e.g. WestLaw, Practical Law, PACER etc.). It requires tremendous effort merely to search for knowledge across these multiple data stores, leaving little time and resources for the other KM activities.

This finding is reinforced by the results of a recent study on KM conducted by Thomson Reuters... seamless knowledge retrieval experience across the different data sources." with "This insight is reinforced by our finding from a recent workshop conducted by Thomson Reuters. As part of our incubation program with Thomson Reuters Labs, we participated in the Design Thinking workshop with Knowledge Managers from various firms. The participants converged on the desire to have their data integrated on a single interface, with a consistent way of classification and a seamless knowledge retrieval experience across the different data sources

From a technology perspective, it is neither difficult nor novel to build an aggregated platform which connects to different sources of data. Many technology companies have tried to address this via enterprise search or federated search solutions. The real difficulty lies in how these disparate sources of content would be indexed for the purposes of ranking the results and providing a unified search experience thereafter. If the data in question is similar in nature but merely comes from different sources (think travel fare aggregators, mortgage broker websites), it is possible to apply the same index to achieve satisfactory results. However, if the types of data are different — which is the case with knowledge required by a lawyer— the index will have to be sufficiently generic to be applied to all types of data. Following this to its logical conclusion, a data aggregator platform for legal KM will work well only for the lowest common denominator of all types of data, i.e., keywords.

Unfortunately, keyword search is insufficiently precise for lawyers. It is unable to distinguish between the different usage that a word may have. For example, “consideration” is a legal term of art under contract law and sees popular usage with its everyday meanings in legal documents (“policy consideration” or “consideration of mitigating factors”). Further, keyword search requires Boolean operators in order to refine a multi-word search query. A lawyer will have to break down a search on shareholder disagreement into “deadlock” AND “shareholder” AND “Russian roulette” so that the search does not return results regarding locking a shareholder in Russia!

Technological progress made over the last 15 years have made people realise that search augmented by artificial intelligence can help them surface better results for their search. This can mean a single platform aggregating knowledge from disparate sources, organised and ranked in ways which would facilitate discovery by users.

The Cure: of metadata, taxonomies, knowledge graph, and machine learning

On the question of how knowledge from different silos should be integrated, we are a firm believer that “the first step to sanity is filtering - filtering the information to extract for knowledge. Filter first for substance. Filter second for significance [5].”

The “substantial” and “significant” information which we are interested in may be informed by a document’s metadata, which is data about the document (or file, if the document exists electronically). More specifically, the metadata can describe the contents of the document. On an electronic system, the metadata would allow us to filter and automatically classify documents into dimensions consistent with the mental model which lawyers have when they are recalling knowledge documents. The dimensions mentioned may be structured and guided by different taxonomies, but can be made to align for standardisation.

This might sound complicated, but it actually happens on a regular basis in a law firm. Consider this example: After reading a commentary, you determine that it has answered questions about transferring shares which may be relevant to a specific client. You want to retain it on your computer and organise it in a folder representing the “client ID”; Next, you create a sub-folder to indicate the “matter number”; within the “matter number” folder you further distinguish, via more sub-folders, the different types of document that you will prepare for this client; finally, you create even more sub-folders within “client advice” to represent the various potential legal issues relating to this matter. You’re happy with this organization and save all your future files accordingly. By following this organization system, you have intuitively created a system of taxonomies and a structure to assemble these taxonomies.

The great tragedy of this story is that such file administration is crucial for future knowledge discovery, by you or your colleagues, yet extremely time-consuming and unbillable for lawyers.

But the heavy lifting of organization and classification according to defined taxonomies may be done by pre-trained machine learning models. Hours of tedious KM work can be reduced to minutes, and it avoids the problem of having inconsistent taxonomies.

Another thing to note in the example above is that the human brain is remarkably able at establishing context when reviewing information. However, as good as humans are at creating connections between unstructured data and establishing context, we are extremely constrained in the amount of data we can monitor and evaluate [6]. Technology can assist humans in overcoming this obstacle, by plotting this mass amount of data onto a “graph”. A “knowledge graph” makes use of description logic to represent a collection of interlinked entities relevant to a specific domain. The knowledge graph does not just list entities, but also represents their interdependent properties and relations. Especially relevant to the legal industry is that knowledge graphs can also contain entities like documents [7], from the different data sources and also a law firm’s internal know-how being represented side-by-side.

Most importantly, artificial intelligence, as applied on knowledge graphs, allows inferences through the relations and assertions relating to the entities. An example would be if “floating charge” IS A “charge” && “charge” IS A “security” So a lawyer who searches for “security” will see results relating to “charge” and “floating charge”. With the use of a knowledge graph to power search, it is thus possible to uncover links between documents from different data sources and allows for a more comprehensive way for lawyers to retrieve relevant knowledge.

Closing remarks The competitive landscape in the legal services industry is rapidly adapting to its clients’ expectation to deliver more with less. In this Information Age, all law firms face the challenge of information overflowing from disparate sources. Law firms that learn to harness the power of modern technologies in their design of a comprehensive KM strategy will be well positioned to turn this challenge into an opportunity. With the right technology, process and culture in place, we are confident that the KM imperative will help law firms achieve better business results in this new age.


[1] The Digital Business Law Group. “Search, Knowledge Management, and the Practice of Law”. 2009

[2] AdvanceLaw. “GC Thought Leaders Experiment”. 2018

[3] Matthew Parsons. “Effective Knowledge Management for Law Firms”. 2004

[4] The Digital Business Law Group. “Search, Knowledge Management, and the Practice of Law”. 2009

[5] Marc Stiegler (1998). Davids Sling. Retrieved from

[6] A Forrester consulting thought leadership paper commissioned by Microsoft. “Extending the value of AI to Knowledge Workers”. 2019

[7] A. Blumauer. “From taxonomies over ontologies to Knowledge Graphs”. July 2014

About the Author


Ellery Sutanto heads business activities at INTELLLEX, evangelizing the strategic importance of Knowledge Management (KM) in the legal industry. In his role, Ellery consults with different stakeholders at global law firms to help them understand how technology might be implemented to actualize their KM strategy.

Ellery is one of the co-founders of INTELLLEX, a Singapore-based company that provides artificial intelligence powered KM solutions for legal professionals.

Ellery van be contacted at

bottom of page