The rapid integration of transformer-based “AI” models into enterprise and consumer workflows has been remarkable. Github copilot has 1M paying users across 37,000 organizations, and Adobe Firefly’s initial model has generated over 3 billion images in less than a year. Established incumbents, leveraging their deep customer relationships and distribution channels have experienced almost instantaneous adoption of AI features. In contrast, we see startups finding success by intensely concentrating on vertical problems, aiming to make a specific workflow more efficient and drive superior results.
The legal landscape is one such example. Large Language Models (LLMs) are known for being particularly good at long-range dependencies and capturing contextual information. These systems excel at generating human-readable summarizations, making suggestions based on enormous amounts of ingested information and greatly simplifying repetitive work. Coupled with the legal industry’s requirement for precise language and detailed documentation, it becomes the ideal domain for generative AI to improve information processing tasks. Although historically slow adopters of technology, lawyers are embracing new tools to increase their outputs and/or optimize their information flows.
In our recent whitepaper, we emphasized that measurable and sustainable customer outcomes require a process of thoughtful experimentation. A detailed understanding of a user’s workflow, speed, and agility leads to an outsized potential to inject new value into crucial workflows. This is why we are thrilled to announce that we recently led Spellbook’s US$20M financing round. Spellbook is an AI copilot for lawyers used by over 1,700 law firms worldwide. Spellbook supports its customers with the resources to make more accurate and efficient legal decisions when drafting contracts and supporting documents.
We anchored our thesis in Spellbook on the team’s capacity to iterate and capitalize on the current AI wave. Years before LLMs became the default accelerating technology they are today, Scott and his team had been working with lawyers to help them automate the drafting of routine contracts using advanced templates. In 2019, Spellbook first started experimenting with GPT2, years before ChatGPT took the world by storm. Their ability to remain nimble, leverage technology and their pervasive knowledge of the customer gave Spellbook an advantage when transformer models finally became “good enough.”
This experimental DNA, paired with timing, superior customer experience, and a commitment to measuring results, helped Spellbook grow revenue 10x from 2022 to 2023. Their user base has increased nearly 300%, and more than 86,000 contracts are being opened on the platform monthly. Lawyers are saving 30% of their day by using Spellbook, and the team continues to see daily usage and retention increase, propelling them forward in their mission to scale into 30,000 law firms worldwide.
As we look forward, it’s clear that Spellbook is not just riding the wave of AI super-powered workflows in legal tech – it’s steering it. Together, we see a future where legal expertise meets technological innovation and redefines what’s possible.
If you’re an entrepreneur redefining your industry – we want to hear about it.