Law firm leaders considering generative AI tools now have to account for them in the technology budget, and some providers' pricing may deviate from the traditional software standards of paying per user, for data, or both.
"The investment could be very large, potentially," Andrew Sprogis, chief innovation officer for Katten Muchin Rosenman LLP, told Law360 Pulse. "We want to make sure if we're going to do anything, it's for the right reasons."
What Is the Pricing?
While law firm leaders who spoke to Law360 Pulse declined to disclose specific or estimated costs for their firms, Law360 Pulse reached out to some legal companies with AI chatbots to understand more about the pricing.
Legalyze.ai, a startup that helps lawyers query case files by using AI, currently prices its tool at a base plan of $299 per user each month, with no long-term contracts and unlimited use with the newest version of OpenAI's language model, GPT-4. There's also an a la carte plan of 75 cents for each query.
"Some legal tech companies in AI take advantage of lawyers by charging high prices for single document AI generation, such as $500-plus for the generation of a demand package," Chris Ford, the founder of Legalyze.ai, told Law360 Pulse. A demand package refers to a demand letter and exhibits, including those used in personal injury cases.
Casetext, which Thomson Reuters acquired for $650 million in June, prices its AI legal assistant CoCounsel at $400 per month for an annual subscription or $500 per month for monthly access. The company also offers one-time access to a single skill, or use, for $50 to $100 depending on the skill needed.
The legal assistant Paxton AI is currently free but plans are to charge $99 to $500 per month for each user.
The generative AI tool Alexi currently offers a free, 14-day trial plan with two memos and 20 arguments. Its $499 per month plan includes five memos and 50 arguments per month, and its $949 per month plan includes 10 memos and 100 arguments per month.
LokiBots, an AI platform used for legal and other functions, has plans that range from $500 to $2,000 per month.
The AI legal research platform vLex told Law360 Pulse that its pricing strategy isn't public yet.
Lega, the AI governance startup for law firms, declined to make its detailed pricing available at this point. CEO Christian Lang told Law360 Pulse that the company is still working with customers on its long-term pricing model, but "we believe strongly that enterprises shouldn't have to spend hundreds of thousands of dollars to begin exploring generative AI."
LexisNexis Legal & Professional, the parent company of Law360, declined to disclose pricing for its Lexis+ AI tool.
The legal tech startup Harvey, which raised $21 million in April, did not respond to a request to comment on its pricing.
Where Law Firms Stand
Sprogis said Katten is currently exploring the different generative legal AI tools on the market, and plans to potentially test a tool by the end of 2023.
Katten declined to say which AI tools it was considering because the firm was still in the middle of negotiations with different providers as of September.
Still, Sprogis said the obvious about the tools: "They're not cheap."
Generative AI as a product category in legal is still very new, with most current tools in the beta testing phase. This leaves law firms scrambling to assess the potential cost versus the value of how it fits into client services.
Tech leaders from Troutman Pepper, which built an internal AI assistant and is also currently evaluating AI providers, told Law360 Pulse that vendors are still sorting out their pricing models while trying to determine what the market will tolerate in terms of pricing.
Eighty-six percent of law firms predict an increase in legal tech spending in the next three years, according to a survey done in summer 2022 and released by Meritas, the global alliance of independent law firms, in December. But that survey was conducted months before OpenAI launched ChatGPT in November and the age of generative AI in the legal field began.
Sprogis said the price tag for some generative AI tools is high, but some tools are less expensive than others, which may not necessarily increase the overall tech budget for the firm.
Determining the return on investment for a generative AI tool depends on the use case, he said.
"Without knowing the use cases, it's hard to figure out why we should make the investment," he said.
Sprogis said that leadership supports the efforts to explore generative AI, and that the budget is flexible to accommodate innovation.
Troutman Pepper's team is still trying to determine the potential costs of using other generative AI tools.
Partner Alison A. Grounds told Law360 Pulse, "I don't know that we have the answers to all that yet because I don't think the providers have the answers to all that yet."
Tokenized AI
Complicating the pricing is that some providers may charge firms based on the length of a query, which can vary per use.
Microsoft's Azure OpenAI Service, which several law firms have used as the model for custom-built generative AI chatbots, prices its language models based on tokens.
A token represents a part of a word, such a handful of characters, or punctuation marks — so an entire sentence could cost about a dozen tokens, depending on the length of that sentence.
As of September, Microsoft charges anywhere from 0.15 cent to 12 cents per 1,000 tokens, depending on the model used and whether it is for a prompt or a completed output from the tool.
Legal briefs could contain tens of thousands of words, and each word might cost one to two tokens.
If the pricing sounds complicated to predict, well, that's exactly what some law firms are dealing with right now.
The Australia-based firm Allens uses the language models from Microsoft for its newly built generative AI chatbot.
Bill Tanner, chief information officer for Allens, said that keeping track of the token costs is a challenge, but that the firm added some flexibility to its tech budget because of that uncertainty.
In theory, Tanner said, using more tokens on a prompt could result in time saved on complicated legal tasks.
"If you're getting more value out of it because you've used more tokens, then that's a great story," Tanner said.
However, he said it's difficult to track the time saved right now in order to prove that value.
While the cost per token is low, the number of tokens can really add up in the legal field, where lawyers may ask a generative AI tool to review hundreds of thousands of documents.
Troutman Pepper's AI assistant is also built with the models from Microsoft.
William Gaus, chief innovation officer for the firm, said it has had discussions about the future of its chatbot, recognizing that at some point the firm may have to limit the use of the tool to certain staff members or for complex use cases if the cost adds up.
Pricing questions aside, law firm leaders say they are looking forward to learning more about the capabilities of these new generative AI platforms to see how they can be used to help clients.
"We hope and are excited about a time where we will be able to leverage that technology to reduce the cost of providing legal services and continue to improve the quality," said Grounds of Troutman Pepper.
--Additional reporting by Matt Perez and Sarah Martinson. Editing by Marygrace Anderson.
For a reprint of this article, please contact reprints@law360.com.
