Significance of AI in Legal Ops and contracting
Q. What are your thoughts on the ever-increasing significance of AI (not just GenAI) in legal services and the way in which it is shifting the broader legal technology conversation?
Aravind: “In the legal space, AI helps identify risks during the drafting and negotiation phases. While AI isn’t advanced enough to replace human experts or lawyers, it serves as a valuable tool for early risk detection. Our journey began with digitising documents and recognising clauses and metadata through our auto extraction (AE) tool. Now, we’re advancing further by analysing contract deviations during negotiations, and comparing them against an organisation’s preferred positions or playbook. These are some of the key challenges we’re currently addressing.”
AI strategy and examples of implementation
Q. Talk to us specifically about Sirion – what is Sirion doing in the AI space?
Aravind: “Sirion offers a comprehensive Contract Lifecycle Management (CLM) platform that focuses on creating, storing and managing contracts. The platform integrates AI and generative AI (Gen AI) to enhance various stages of the contract process:
- In the Creation phase, AI aids in tasks like metadata extraction and identifying clause deviations, while Gen AI helps with issue detection and automated redlining during negotiations.
- For Storage, the platform now features a conversational search powered by Gen AI, allowing natural language queries across all documents.
- In the Management phase, AI assists with obligation extraction and financial data management, with future plans to expand Gen AI use cases further.”
Q. I understand it’s still early days, but do you have any examples of how AI has been applied in real-life use cases?
Aravind: “Yes, we are actively rolling out the AI use cases with several live customers.
The first example is out Turbo Auto Extraction. Previously, we relied on small and medium language models, but they struggled with extracting interpretive fields like Effective or Expiration Date, where the relevant data is scattered across the document. Now, we’re using Large Language Models to improve accuracy by training them with the right prompts. This enhanced extraction capability is already in production and being used by various customers.
The second example is our conversational search feature, which allows users to query their entire document corpus using natural language. Powered by LLMs, this feature translates open-ended questions into precise queries, fetching highly accurate data from both metadata and full text. This is a significant improvement over legacy systems, where users needed to know exactly what metadata fields to search for. Now, even non-legal experts can ask complex questions, like assessing exposure to a particular risk, and receive intelligent, helpful answers.”
Best practices of AI implementation in CLM
Q. What’s your approach to AI implementation in the context of CLM?
Aravind: “Like any other software, we invite customers to our early access program. This allows them to test the product and provide feedback on its accuracy and usability. Customers can also give real-time feedback within the product, similar to a thumbs-up or thumbs-down system of ChatGPT. We engage with both current and prospective customers through this program and our annual Customer Advisory Board. Once the product meets our standards and receives positive feedback, we proceed with broader rollout. This early feedback is crucial, as it helps us refine and improve our AI models beyond the initial 80& accuracy achieved through internal testing.”
Q. In terms of the roll-out of the solution for actual client projects, what are your best practises?
Aravind: “AI is a powerful tool, but it’s not yet mature enough to fully replace a knowledge expert. Like a calculator or spreadsheet, AI serves as enabler, boosting solutions across various use cases. However, it’s not 100% accurate of free from errors, and it works best when combined with human expertise. The technology is evolving rapidly, and we must continual update our models to improve accuracy. Nonetheless, there are limits to how precise AI can be compared to an expert lawyer, and it’s important to recognise that AI is not a one-size-fits-all solution.”
Q. It’s really interesting that you mentioned the human element. This brings me to my next question – how does Sirion’s relationship with DWF enhance AI implementation?
Aravind: “In my view, there are significant gaps in what AI can achieve compared to human expertise. For nearly every use case we’ve encountered, whether it’s digitisation, conversational search, issue identification or redlining, there’s a crucial synergy between human and machine. DWF provides not only legal expertise but also project management expertise which are essential in implementing AI technologies. The combination of legal tech professionals with AI tools really creates valuable synergies that enhance the overall effectiveness.”
Future of AI
Q. After all the talk about what AI can do, what is your perspective on how legal professionals are going to have to change to adapt?
Aravind: “In my opinion, AI will transform the legal profession by automation routine tasks that many of us would prefer to avoid, freeing up time and energy for more complex and creative work. While AI can handle repetitive and predictable tasks, it’s not a stage where it will replace entire jobs, especially those involving creativity. Instead, it will enhance productivity, allowing both junior and senior lawyers to accomplish more, but it won’t make legal professional obsolete.”
If your team would like to learn more about how to best utilise AI tools in the legal profession, contact our team below.
