Typically, firms reach out to vendors with particular use cases in mind. But not with Harvey, the legal tech startup that raised over $200 million to roll out its generative AI assistant to law firms.
Generative AI tools can create original content based on data they were trained on.
Esther D. Bowers, Honigman's chief practice innovation officer, told Law360 Pulse that it "is a bit unique" that the vendor reached out to the firm last summer, contacting a group of lawyers with the firm who blog about data privacy.
"They knew that we already had some forward-thinking individuals in our firm as it relates to the adoption of these sorts of technologies," Bowers said.
Internal discussion at Honigman about Harvey began shortly after the vendor reached out.
Honigman's practice technology committee, a group of about a dozen individuals who represent different practices at the firm, evaluated the tool to see how it might be used within the firm's privacy policies. The group then mapped out how to potentially use the tool and what the rollout might look like.
This process took a few months because the Mid-Law firm had to first determine the pain points that Harvey could potentially solve. The firm is finding emerging use cases within its patent, mergers and acquisitions, and private equity practices.
Bowers said this process was "completely unique" to Harvey because those use cases weren't fully clear at first.
To find those use cases, Honigman worked with Harvey to explore examples of how other firms have used the tool. From them, the firm evaluated how those use cases fitted within Honigman's current processes and workflows.
This resulted in weeks of further discussions surrounding the potential benefits of the tool. Discussions included how it could impact the firm's ability to build the talent pipeline and how it could be used to improve the firm's retention strategy.
To fully understand all these points, the firm needed to pilot the software.
"You really can't experience that until you're in a pilot or you can really play with the tool and understand what it can do," Bowers said.
In early 2024, Honigman officially trialed Harvey.
About 30 Honigman attorneys piloted Harvey around February and March 2024. Nonpracticing attorneys also participated in the pilot because they are often responsible for helping in various projects and need to understand the capabilities of the tool.
Bowers tested Harvey by challenging the platform to surface information about a business or summarize information.
"Harvey's ability to access the EDGAR database was really helpful," Bowers said, referring to the Securities and Exchange Commission filing system.
Honigman then tested the AI assistant CoCounsel from Casetext, which was acquired by Thomson Reuters, to compare it to Harvey.
Bowers said that both tools are unique in their capabilities, which is why Honigman signed agreements with both providers around June. The firm said that Harvey will be used for analyzing contracts and conducting due diligence.
Honigman started rolling out Harvey across the firm in July. The rollout for CoCounsel has not begun.
The firm requires users to complete ethics training before gaining access to Harvey. Following ethics training, users then get training for Harvey, which is provided by the vendor.
Honigman is organizing the staff into small groups for the training, which will be conducted on a biweekly basis and be used to capture ideas and use cases. There are also plans to invite Harvey representatives to those smaller group training sessions to help with troubleshooting and provide more guidance on the technology.
Bigger Picture
There's no question that Harvey is one of the most hyped legal tech startups in recent years, but the technology has been kept under wraps. Until recently, Harvey did not disclose information about its suite of products on its website.
Despite all of its funding, only about 1% of law firms are using Harvey, according to a survey from Law360 Pulse in April.
Bowers said that Harvey's platform has evolved since last summer with "massive improvements."
"The [user interface] is very clean and easy," Bowers said. "It's really easy to understand how to navigate within."
Pricing for legal generative AI tools can be exorbitant, including for pilots, according to experts. Bowers said that pricing for these types of tools during the pilot can vary depending on integrations with the firm's existing tech stack.
Honigman measures return on investment for Harvey and other AI tools by looking at associate retention, development and overall satisfaction.
"We have run parallel assignments testing the technology to compare output, efficiency gains, accuracy and quality with an associate functioning with the tool versus without it, and have found these metrics typically increase," Bowers said. "These tests enable our firm to make determinations on how we use AI technology as an assistant — not a replacement — focusing on the value a human can provide when the rote tasks can be performed by these tools."
Another important benefit is that lateral candidates are more attracted to firms that adopt and promote AI platforms, according to Bowers.
For other law firms looking at generative AI tools, Bowers recommends first understanding the purpose of the tools. Bowers said firms should be accepting of errors during the journey while being flexible enough to work with a tool in new ways.
"I continue to talk about our need to be very patient through this process because it is new, it is rapidly evolving," Bowers said.
Bowers added that a law firm's decision to use generative AI should be about improving legal services.
"This is really about us deepening client relationships and automating and allowing technology to do the tasks that it can do, so our attorneys can be focused on deepening those relationships and being trusted advisers to our clients," Bowers said.
--Editing by Robert Rudinger.
For a reprint of this article, please contact reprints@law360.com.
