Spellbook Launches Gen AI Agent that Can Plan and Execute Complex Transactional Workflows

Spellbook, the company that introduced the first generative AI copilot for contract drafting and review in 2022, today launched what it says is the first AI agent for law, capable of planning and executing complex, multi-step workflows.

Called Spellbook Associate, the agent can plan, execute and check its work, and adapt to accomplish larger scope assignments, much as a junior associate would, Spellbook says.

“We believe that agents are the most important leap forward in AI since ChatGPT, and that they’ll be even more impactful,” Spellbook cofounder and CEO Scott Stevenson said. “Chat tools were helpful for answering one-off questions, but soon we’ll have AI colleagues in every industry who will be working through complex deliverables, similar to humans.”

Whether this is actually the first AI agent in law is difficult to determine, as there is not even a common definition of what an AI agent is. A quick search finds that Bryter announced an AI agent last month.

But if we think of an AI agent as an LLM-driven application that can autonomously perform multi-step complex tasks, then, whether first or not, Spellbook Associate is impressive, based on what I have seen.

(For two useful overviews of AI agents in law, see this Law.com article and this Law.com article.)

A Demonstration of Associate

In this video, Spellbook cofounder Daniel Di Maria, a former lawyer and now the company’s chief revenue officer, shows an example of Spellbook Associate executing a multi-stage workflow involving a financing transaction.

Di Maria uploads a term sheet and asks Associate to use that to update four other standard documents. Associate initially responds by outlining the workflow plan, describing what it will do to execute the plan, and asking if it should proceed.

Di Maria approves it to proceed and it then summarizes the term sheet, again giving him the option to review the summary and give the OK to proceed. (An auto mode goes through the workflow without asking for approval at each step, but everything can be reviewed at the end.)

Image courtesy of Spellbook.

From there, Associate creates the editing tasks, and Di Maria is able to separately execute each task by clicking play. That causes Associate to open the document in Microsoft Word and begin suggesting changes across the document. Di Maria has the option to review each change or apply them all and then review them using Word’s track changes feature.

Associate includes both a rewind feature and versioning, so a user can always go back to any point in the session or to any prior version. Nothing it does is permanent and it creates a paper trail of everything.

The company says that Spellbook Associate executes projects in a fraction of the time junior associates require, allowing lawyers to focus on higher-value activities such as negotiation and building client relationships. It can also enable small to medium-sized law firms to more effectively compete for opportunities typically reserved for larger firms.

“Just as AI agents have been rapidly adopted across software engineering over the past six months, we believe they’re coming to knowledge workers in every industry,” Stevenson said. “We’re proud to be the first to bring a true ‘AI Colleague’ to lawyers.”

(Note: Spellbook CEO Stevenson will be my guest on the next LawNext podcast, which will be out Tuesday. To catch it when it comes out, subscribe to LawNext through your favorite podcast player, or watch for it here.)