Jay is the Founder & CEO of Eve. He has over 15 years of experience in AI and Machine Learning.
Prior to Eve, Jay was an early stage investor at Lightspeed Enterprise Partners. Before that, he was the First Engineer and Head of Product Engineering at Rubrik, helping construct the corporate from the bottom up.
Eve is the primary personalized legal AI tool built for the legal occupation you could partner with, train, and teach, just like several other member of your team.
What initially attracted you to engineering and computer science?
Once I was heading to varsity, I originally planned to change into a chemist/material scientist. But I quickly found the pace of progress with computer science was several orders of magnitude quicker than anything on offer. This felt like a momentous occasion, a time in history where we had the chance to make use of our brain power to enhance virtually all the pieces we do. I used to be hooked.
But taking a step back, I even have been a “builder” since I used to be a young kid. ’ I used to spend a whole lot of time constructing things out of toys and other small odds and ends. My crowning achievement: a toy elevator made out of a bear-shaped jello container that will ferry broken toys up and down on vital toy business).
My passion for constructing things never went away, I just graduated from broken toys to artificial intelligence and am at all times looking forward—what can we construct next?
You were First Engineer & Head of Product Engineering at Rubrik for over 4 years, what were you working on and what did you learn from this experience?
Over my time there, I had the tremendous opportunity to grow with the people around me. From a technical perspective, I worked on all the pieces from a completely distributed job scheduling framework to the front end code, and later progressed into management.
The most useful experience that I got while working at Rubrik was the experience of constructing a very transformational product, with many lived and learned examples of what it takes to achieve this.
After this you became an early stage investor with Lightspeed Enterprise Partners, what kind of AI corporations were in your radar and what qualities did you search for in a team when investing in startups?
I used to be an Enterprise AI investor, primarily on the lookout for corporations enabling others to construct with AI. I primarily searched for clarity of thought on the important thing pain point they were solving, and why they thought their framing of the issue was the correct one. Beyond that, my top must-have was the flexibility of the founders to execute the answer.
What made you realize that you just desired to return to your engineering roots, and launch a legal startup?
I’ve at all times been obsessed with constructing. Even during my time as a enterprise capitalist, I’d create sample apps to validate my ideas. This hands-on approach was particularly enlightening when exploring AI corporations. A matter began to captivate me: how can we create a world where AI empowers people? This growing fascination led to the inception of Eve. I selected the legal field as our start line, not only for its complexity but since it presented clear, day-to-day challenges that AI could address. Essentially the most thrilling aspect for us is the transformative potential of AI and software in redefining workflows. We’re not only talking about incremental improvements; we’re envisioning a whole overhaul of how law firms operate. This can be a groundbreaking shift, and it’s what excites us essentially the most.
What kind of legal research and legal assistance is Eve able to?
Eve is able to performing legal research—searching through a repository of court rulings updated every day, identifying key case law. Eve uses natural language search to perform legal research, and may incorporate key case details into the search. Eve can also be able to helping brainstorm and evaluate key legal claims and procedures throughout the lifecycle of a case.
How is the info sourced to coach the AI models?
Eve uses essentially the most advanced Generative AI models to power responses, including GPT-4, Anthropic’s Claude 2 and internal proprietary AI models. Eve’s foundation models were trained on publicly available information on the web, in addition to large legal datasets provided by Eve’s partners. Eve leverages a state-of-the-art LLM Ops framework to be sure that our models are continuously evaluated for accuracy and up-to-date-ness. As recent innovative models (corresponding to GTP-4 Turbo and Claude 2.1) are released, we work through a diligent evaluation, analyzing them for quality after which add them to our framework.
Laws and regulations change rapidly, how does the AI keep current with the newest news and developments?
An important a part of the model that should be updated with recent laws, statutes, and court rulings is the Legal Research functions. Our Legal Research feature is built with a RAG approach —and this product sources information in a different way than the remainder of the model—partnering with an open-source court case reporting provider that updates its database every day. This permits us to have legal information that’s up-to-date, surfacing key rulings and decisions for legal professionals.
Hallucinations is one among the core issues with LLMs, what is going on on the backend to mitigate or reduce these potential issues?
We’re using multiple approaches to assist prevent hallucination. The predominant risk that we see for legal is an AI hallucinating and making up a legal case, presenting it as a cite-able resource. We’ve taken extra care to create a legal research functionality for Eve that employs the retrieval-augmenter-generation (RAG) method for looking for relevant case law. When performing Legal Research, Eve is looking right into a closed box of court rulings—unable to make up a case that isn’t in our database. Outside of legal research we’re working diligently to refine our models. We also put concerted effort into enabling our users with pre-built and curated “skills” that they will use, that are prompt-engineered to provide optimal results. Our qualified success team does white-glove onboarding and support, helping to teach users on learn how to use generative AI within the safest way possible.
What’s your vision for the long run of AI within the legal occupation?
AI is beginning to disrupt the legal landscape, and we predict it is barely starting its journey. The advances in AI enable knowledge employees corresponding to legal professionals to attain superhuman capabilities. Our mission is to empower legal professionals to harness the facility of this recent technology and augment their very own abilities. We envision a world where AI is an integral a part of all case work. Our goal is to construct a product that not only helps increase lawyers’ efficiency, but starts to work alongside legal professionals, considering and reasoning alongside them—helping with increasingly useful and strategic parts of their workflows.