Novelty? Or practical technology? That was the question on people’s minds when generative artificial intelligence (AI) first emerged.
Just a few short years later, the use cases for AI in the legal realm are starting to snowball as it helps with a growing number of tasks — and puts to rest any lingering question around AI’s utility for legal administrators.
Say that a corporate human resources user is tasked with looking at the last six months of employee agreements and reporting on salary trends and whether they include a termination clause. The user can run a search for “employee AND agreement,” then filter by appropriate library, and then narrow by matter/workspace and date range. This will get them a collection of documents. However, that’s just the start of the task, not the end of it. Completing the assignment means reviewing them one by one.
Instead, the user can use a generative AI-style chatbot to interact with those documents by asking, “What is the salary?” “Is there a termination clause?” and “Summarize the termination clause.” The generative AI, in other words, takes over the burden of having to review the corporate employee agreements and find the relevant information.
From there, it’s easy for the HR professional to export the results to Excel and prepare a report on what the findings show. The key element here? AI has enabled them to spend more time on the high-value aspect of the task — interpretation and analysis of the findings — rather than the grunt work of gathering the data.
Let’s take the case of the support team within a corporate legal department. This team deals with hundreds of contract queries every week and is overloaded with work. They might have a ticketing system in place to receive those queries, but all requests for contract review are manually triaged, directed and dealt with.
This is a prime area for AI to lend a hand and streamline the process. A generative AI tool can extract key information from the request and hand it off to the ticketing system. The overloaded corporate legal team no longer has to waste time on manual triage: Work is automatically routed to the appropriate workflow. As a result, they can focus on doing other aspects of their job and responding to their clients faster.
Within organizations, there are often playbooks that serve as essential guidelines for structuring contracts with customers, vendors and other business entities. These playbooks define necessary contract clauses, set parameters and ensure consistency across agreements.
AI can play a crucial role in this context. When drafting a new contract, an intelligent AI agent can continuously cross-reference the contract with the playbook — ideally during the early stages of creation. If any deviations occur, the AI promptly identifies and highlights them.
“Of course, to successfully deploy AI for any of these use cases, there’s some groundwork that needs to be laid ahead of time. For starters, companies should evaluate their information architecture’s current condition.”
The AI can extend its utility even further by assessing an organization’s existing contract portfolio against any updated playbooks. Manually verifying compliance would be a particularly time-consuming effort, especially if there’s a large volume of contracts. AI can perform this task either on an ad hoc basis for specific contracts or by analyzing all contracts within a specific sector to pinpoint deviations from the latest playbook. This practical application of AI significantly streamlines the process for busy professionals.
Of course, to successfully deploy AI for any of these use cases, there’s some groundwork that needs to be laid ahead of time. For starters, companies should evaluate their information architecture’s current condition. Training generative AI’s large language models (LLMs) requires data, raising important questions about the organization’s reliable data sources. More specifically, identifying where these dependable data sets reside is crucial.
While a document management system (DMS) provides a solid base, it’s merely an initial step. The vast array of documents in a firm’s DMS may contain too much “noise,” making it challenging for generative AI to discern the valuable “signal.” It is more effective to introduce LLMs to a select portion of data — for instance, only the definitive versions of documents from a particular period, rather than every version dating back indefinitely.
However, technology is not the sole solution. It’s essential to have dedicated personnel and processes to select the “optimal” data for training the model or validating the results. Establishing a robust knowledge management and curation framework within the company ensures the continuous upkeep of the training data set, rather than treating it as a sporadic task. Once this groundwork is in place, organizations are well positioned to start reaping the benefits of AI across a rapidly growing number of use cases.
We are only just starting to scratch the surface of ways to put AI to work, which means that there are plenty of use cases and benefits waiting to be discovered to help legal organizations deliver better business outcomes.
