Scott Stevenson, is Co-Founder & CEO of Spellbook, a tool to automate legal work that’s built on OpenAI’s GPT-4 and other large language models (LLMs). It has been trained on an enormous dataset of 42 terabytes of text from the Web as an entire, contracts, books and Wikipedia. Spellbook is further tuning the model using proprietary legal datasets.
What initially attracted you to computer engineering?
I loved video games as a child, and was inspired to learn how you can make them as a teen–that set me on the course of becoming a software engineer. I’m drawn to the career’s inherent creativity and in addition appreciate the hardware aspect intertwined in computer engineering.
Are you able to discuss how your experience with GitHub Copilot was the initial inspiration for Spellbook?
We had been working with lawyers for years, attempting to help them automate the drafting of routine contracts using advanced templates. They’d often say the identical thing: “templates are great, but my work is simply too bespoke for them.”
GitHub Copilot was the primary generative AI assistant for software engineers–you’ll be able to start writing code and it can “think ahead” of you, suggesting large chunks of code that you just might want to jot down next. We immediately saw how this might help lawyers draft bespoke agreements, while also helping them intelligently “auto-complete” contracts.
How does Spellbook suggest language for legal contracts?
In the primary version of our product, we offered a classy auto-complete feature, just like Github Copilot. Now we have now a variety of other mechanisms:
- Spellbook Reviews can take an instruction like “aggressively negotiate this agreement for my client” and suggest changes across a whole agreement.
- Spellbook Insights routinely finds risks and suggested clauses across an agreement.
Spellbook also reviews contracts, what sort of insight does it offer legal professionals?
Spellbook offers a wide range of insights during contract reviews for legal professionals. These insights might be tailored using different “Lenses.” We offer default lenses for tasks like contract negotiations, but lawyers may provide custom instructions, equivalent to “Review this contract to make sure it complies with California customer requirements.”
Spellbook can uncover potential risks, discover oversights, pinpoint inconsistencies, and receive useful suggestions for improving and enhancing contracts.
Are you able to describe how Spellbook overcomes the token size limits which are offered by LLMs?
It is a significant a part of what sets us apart and constitutes our unique approach. Managing lengthy contracts that might be greater than lots of of pages can put a strain on an attorney’s bandwidth, but Spellbook’s technology excels in handling them efficiently. While we cannot delve into the specifics of our methods in the mean time, that is where our expertise truly shines.
How is the info sourced to coach the AI models?
We’ve availed of public datasets like EDGAR, in addition to proprietary contract data sets we built during our company’s first phase at www.rallylegal.com. Nevertheless, we expect that RAG-based approaches are the most effective approach to incorporate accurate legal data into generated text. RAG allows many data sources, equivalent to a client’s own documents, to be referenced.
Laws and regulations change rapidly, how does the AI keep current with the newest news and developments?
We’re finding that retrieval-augmented generation (RAG) approaches are extremely effective for this. We expect of language models more as a “human reasoning” technology. We generally shouldn’t treat LLMs as “databases”, and as a substitute allow them to retrieve reliable information from trusted sources.
How does Spellbook mitigate or reduce AI hallucinations?
We’ve relentlessly tuned every feature in Spellbook to offer the most effective results for lawyers. As mentioned above, RAG also helps keep results relevant and up-to-date. Lastly, our approach to AI known as “Assistive AI”: we all the time keep the lawyer in the motive force’s seat, they usually have to review any suggestions before they’re acted upon. That is central to all the pieces we do.
For the time being contract drafting and review is the first use case, what are some additional use cases that Spellbook plans on offering?
We’re quite focused on being the most effective tool for industrial/contracting lawyers immediately. One natural extension of that helps lawyers with legal diligence during a fancy transaction. Often law firms will put together a deal room containing every substantial legal document in a company, reviewing for risks and discrepancies across the corpus. Spellbook is working towards implementing this use case!
What’s your vision for the long run of AI within the legal career?
Our “Assistive AI” vision is for each lawyer to have an “electric bicycle” which helps them do their job much faster while producing higher quality work and spending more time on adding strategic value to clients relatively than copying and pasting. We expect AI should come to lawyers and be a “wind at their back” without requiring much habit change. We expect every lawyer will soon have an AI switched “on” during every hour of their work, whether or not they are in Word, emailing or in a client meeting.
This ultimately implies that the 70% of potential legal clients, who cannot afford legal services, will finally have the opportunity to be serviced. We’re really enthusiastic about that too.
