Why Vertical LLM Agents Are The New $1 Billion SaaS Opportunities

As large language models (LLMs) continue to evolve, the potential for vertical AI agents to transform industries is becoming increasingly clear. In this episode of Lightcone, hosts Gary, Jared, and Diana chat with Jake Heller, co-founder and CEO of Casetext, about his journey in building a successful vertical AI company and the challenges faced along the way.

Key Takeaways #

• Vertical AI agents are emerging as significant players in the SaaS market.
• The legal industry is ripe for disruption through AI technology.
• Building a successful AI product requires deep understanding and adaptation to user needs.
• The transition from traditional methods to AI can be met with skepticism but can lead to substantial improvements.

The Rise of Vertical AI Agents #

Jake Heller’s journey with Casetext began over a decade ago, focusing on improving legal technology. Initially, the company struggled to gain traction, but the advent of advanced LLMs like GPT-4 changed everything. Heller describes the moment they realized the potential of AI in law as a turning point. The entire company pivoted to focus on developing a new product called Co-Counsel, which leverages GPT-4 technology to assist lawyers in their work.

The Challenges of Law and AI #

The legal field has unique challenges when it comes to integrating AI. Lawyers are often resistant to change, especially when their current methods are profitable. Heller notes that many lawyers were initially skeptical about adopting new technology, fearing it might disrupt their successful practices. However, the release of ChatGPT and similar tools made it clear that AI could significantly enhance their work.

Finding Product-Market Fit #

Finding product-market fit in the legal sector was a long and arduous process for Casetext. Heller emphasizes that the journey involved a lot of trial and error. They initially attempted to create a user-generated content platform for legal research, but this approach failed. Instead, they shifted focus to developing AI-driven solutions that could automate and streamline legal research processes.

Entering Deep Founder Mode #

Heller describes a pivotal moment when he had to convince his team to embrace the new direction. With a company of 120 people, he made the bold decision to shift everyone’s focus to the new AI product. This required a deep commitment from the entire team, and Heller led by example, even building the first version of Co-Counsel himself. This hands-on approach helped to rally the team around the new vision.

The Importance of Prompt Engineering #

A significant part of developing Co-Counsel involved prompt engineering. Heller explains that they had to create a series of tests to ensure the AI could accurately perform legal tasks. This process was akin to test-driven development in software engineering, where they wrote tests to guide the AI’s learning and improve its accuracy.

Going Beyond GPT Wrappers #

Many companies are criticized for merely creating wrappers around existing LLMs without adding real value. Heller argues that true innovation comes from understanding the specific needs of users and building tailored solutions. For Casetext, this meant integrating proprietary legal data and creating a seamless user experience that goes beyond what generic AI tools can offer.

Aiming for 100% Accuracy #

In the legal field, accuracy is paramount. Heller shares that they aimed for 100% accuracy in their AI outputs, knowing that even small mistakes could have significant consequences. This commitment to precision required rigorous testing and continuous improvement of their AI models.

Thoughts on OpenAI’s Capabilities #

With the introduction of OpenAI’s latest model, Heller sees exciting possibilities for the future of AI in law. He believes that advancements in AI will allow for even more sophisticated applications, enabling lawyers to focus on strategic thinking rather than mundane tasks. The potential for AI to enhance legal work is vast, and Heller is optimistic about the future.

Conclusion #

Jake Heller’s journey with Casetext illustrates the transformative power of vertical AI agents in the SaaS landscape. By embracing change and focusing on user needs, companies can unlock new opportunities and drive significant improvements in their industries. As AI technology continues to evolve, the potential for vertical AI agents to create billion-dollar companies is more promising than ever.