Interview with Eve CEO Jay Madheswaran

We sat down with Jay Madheswaran, the CEO and co-founder of Eve, to talk about the present and future of generative AI for the legal industry. 

Q: What are some of the most exciting recent developments in generative AI, and specifically for the legal space?

The biggest recent development is the creation of very large language models like GPT-4 and Claude. These models have been trained on such huge datasets that they’re starting to comprehend language at human-like levels.

What’s particularly exciting for the legal industry is that these massive models are becoming customized, trained on case law and other public secondary legal opinions. We’re seeing these models start to out-perform an average human, for example when taking the bar exam, and they’re going to keep growing more tailored and specialized for the legal profession.

‍

Q: What are some of the biggest opportunities for innovation in legal AI?

One of the biggest opportunities for innovation in legal AI is adding reasoning capability. Current generative AI models like GPT-4 cannot reason - they cannot take learned information and make new logical inferences from it. This type of reasoning is critical in legal applications. For example, when summarizing different aspects of a case and using those conclusions to build a stronger argument. Adding reasoning ability to legal AI would make it capable of assisting with more valuable, complex tasks. This is what we’re building towards at Eve.

‍

Q: What should legal teams consider when evaluating Generative AI tools?

‍Whether you’re just considering adopting a generative AI tool or are actively demoing solutions there is a lot to consider.

There are three things I recommend you look for when you evaluate generative AI solutions. Ask yourself and the vendors these questions: 

1. What are the data security, data retention, and compliance standards? This is table-stakes - any generative AI solution should have deep dedication to data security to match the standard of the legal industry. ‍

‍ - Bonus: Ask how the solution establishes trust in the AI. What protections do they have against hallucination, plagiarism and more. 

2. Does the solution fit into your workflows? Make sure that the solution you evaluate has a way to seamlessly integrate into your expecting process. If there are particular ways that you perform key tasks, or specific preferences on style, tone and formatting - find a tool that can align to your needs and solves your highest ROI problems. 
3. What is the underlying technology and process? There are three kinds of companies in this industry: AI native companies, incumbents adding AI, and upstarts. AI native companies are built from the ground up using AI technology to put it in the hands of legal professionals. Incumbents are embedding AI into existing software stacks which can be limiting. Upstarts are new 1-2 person shops with limited technology stacks. I encourage you to look for AI native companies that have a deeper understanding of the technology, a stronger roadmap, and a more flexible/adaptable product. 

‍

Q: What are some best-practices or recommendations for a firm integrating a new generative AI solution? 

When adopting a new solution, the key is to make the value tangible for all users. This is best demonstrated by a real-world use case - for example showing how the solution would have accelerated a recently closed case. 

The other thing to consider is to map the integration of this tool to your existing adoption strategies. Don’t reinvent how you use new tech, fit the new tech into your tried and true process.  

‍

‍