The AI race is on in legal. Leading contenders (covered in this article) are Macro, Spellbook, Casetext, Harvey, ChatGPT, EvenUp, and Pincites.
As a note, many companies are rushing to incorporate GPT into their products in some way. There are too many to cover, so in this piece, we focus on companies putting LLMs at the center of their products, either as a new company or a new product line.
Innovation breakthroughs don’t come without charlatans. Some companies leverage AI to prop up their offerings without real substance.
Even Open AI’s CEO, Sam Altman, doubts the viability of most ideas funded by the AI frenzy: "It's wildly overhyped in the short-term," Altman said. "There's crazy stuff happening in Silicon Valley right now."
There are, of course, exceptions. In fact, a team of social scientists recently studied the impacts of AI-assisted work at Boston Consulting Group.
They found that for 18 tasks selected to be realistic samples of the kinds of work done at an elite consulting company, consultants using ChatGPT-4 outperformed those who did not, by a lot!
The surprising finding is the emphasis on collaboration between consultants and AI, not the replacement of consultants by AI (which everyone is scared about).
Legal cases require deep expertise. Analyzing contracts often requires cross-referencing complex clauses, provisions, definitions, etc.
General AI models aren’t explicitly trained for legal use cases and underlying analysis. For AI to provide added value to lawyers, it needs to be able to think like one.
At the start of the ChatGPT craze, some lawyers got caught delegating a little too much to AI.
Talk about reputation damages 😅. World-class legal AI must give recommendations without making things up, which can be incredibly risky for lawyers.
Let’s say you’re working on a sensitive case and interacting with an AI model to develop unique defense arguments.
The model stores the sensitive information you provide and uses it for training. A few days later (AI models integrate data fast), the same AI gives your competition an argument recommendation based on your sensitive information.
Not ideal, to say the least.
Each product or company in the legal AI race serves a specific function, which the image below illustrates. Let’s break down each category and explore how each product helps lawyers improve their work.
You may have heard of lawyers delegating work to ChatGPT. Although great for rewriting paragraphs, checking for spelling and grammar, or responding to questions with publicly available information, ChatGPT is dangerous for complex legal use.
If you invested 10 billion dollars into a company, would you integrate its products into yours?
Yes, you would. And that’s what Microsoft is doing with Copilot, fully integrating OpenAI’s ChatGPT product in the Office Suite for an extra $30 per month per user.
For lawyers, Word Copilot is considered more of a productivity tool than a legal tech option. It doesn’t use complex legal doc structure to answer prompts. Like ChatGPT, Copilot spits out general-use answers to complex legal questions.
Obviously, lawyers are expensive, and many in the corporate world consider legal expenses a “necessary evil”.
Tools like Macro, EvenUp, and Pincites can reduce one’s dependency on outsourced counsel by helping in-house teams draft legal documents with the help of AI.
Macro provides AI assistance directly from the editor, which lets lawyers generate new clauses, definitions, or any other type of writing.
It’s a perfect mix between human and AI. Especially considering that Macro’s AI will read through your documents to ensure its generated text is accurate and matches your writing style.
EvenUp is aimed at injury lawyers who want to automate low-value grunt work. It uses AI to write demands for personal injury claims without requiring a legal professional to be involved in the drafting process.
Spellbook is a powerful product built into Microsoft Word. It helps lawyers, mainly small law firms and solo practitioners, with the contract drafting process. For example, people can prompt Spellbook to write a clause, suggest changes, review definitions, or annotate.
Even thought Spellbook is an excellent product, it has limitations. Word-only compatibility renders it unusable on PDFs, a significant part of a document lifecycle.
The buzz around Macro as an all-in-one document workspace for lawyers sparked from its reviewing capabilities like Live Linking.
Lawyers usually scroll or manually set bookmarks to get through long documents. Macro makes documents interactive, so lawyers can click on a section, term, or clause to read the corresponding definition.
Pincites, on the other hand, leverages “playbooks” to automatically generate redlines for legal teams, cutting the review process needed from associates. Like Macro, it has error detection capabilities, spotting potential material mistakes in document drafts.
Here’s an example of error detection in Macro 👇
Casetext helps litigators simplify legal research by asking complex legal questions. The product is well-made and used by thousands of law firms.
It is a standalone web app, meaning lawyers must step out of their regular workflow to benefit from Casetext.
Looking for specific information? Don't waste time searching. Ask your document instead. Macro lets legal professionals search terms or ask AI for contextual documentation analysis.
Harvey has caught mainstream after it raised money from OpenAI’s startup Fund and Sequoia Capital.
Technically, the platform lets you upload legal documents on their website and then ask questions to a chatGPT-like interface. It is an exciting advancement for lawyers who will now have the opportunity to get the benefits of ChatGPT’s Large Language Model (LLM) but with specific legal applications.
We’ve heard from lawyers that Harvey falls short because it does not work well within lawyers’ workflow.
Legal professionals want AI insights as they write or review documents, not in siloes. They want to be able to ask AI a question in between 2 words while hesitating on the governing law, for example.
💡Important reminder: Legal professionals are notorious for change management issues.
Ironclad is an excellent example of leveraging AI to power end-to-end legal workflows in the otherwise unappreciated CLM space.
Ironclad’s AI analyzes contracts, flags areas that require a thorough review, and provides suggestions on negotiating based on legal-approved guidelines.
The problem is - it only makes sense to enable Ironclad’s AI capabilities if your firm is using the full suite of Ironclad’s products; otherwise, it won’t fit in your workflow.
It’s in part because of the Compare feature that Macro’s popularity is rising at lightning speed within the AmLaw 100.
Until recently, there were no viable alternatives to Litera. Macro is the first legal tech product that enables multi-document comparisons, brings a beautiful user interface, fair pricing, responsive customer service, and can run comparisons from email, mobile, and tablet.
👉 Read this article for more information.
Litera is currently the market leader in document comparison software. Its platform helps lawyers compare any two documents together and integrates with the rest of the Litera suite of products.
Doing world-class AI-enabled legal work involves contextual & accurate information + workflow.
By rethinking legal work from first principles, Macro covers all categories without overloading its platform with bells and whistles.
Regarding the technology itself, here are the principles that Macro follows that should be looked for when evaluating legal AI solutions:
Macro GPT uses retrieval augmented generation (RAG) with a large language model pre-trained on legal documents for best performance.
Our cloud is hosted on Azure and each client’s data is partitioned from other clients.
The data from documents uploaded in Macro is inaccessible by LLM providers such as ChatGPT, Bard or Llama.
Macro builds on our academic research conducted at the University of Pennsylvania on ‘Graph Algorithms and Analysis of Complex Legal Documents’. This pre-processing of the document and prompt injection ensures that relevant defined terms, clauses, and references are included in the context window. Further, the output is post-processed to ensure the integrity of the main document. We’ve benchmarked our system against other leading legal GAI providers and are confident in the quality of our results.
Macro integrates with your document management system to pull, edit, and save documents.
Instead of being a separate web app that users have to drag and drop docs into, Macro fits into your existing way of working as your default document editor.
Most teams trying to disrupt legal tech have expert technologists OR experienced legal professionals.
Macro has both. And it shows in how delightful and powerful it is to use the Macro product. 80% of the team are computer scientists or machine learning experts, who thoroughly understand law. Some have worked with complex document workflows at firms like Bridgewater Associates.
We hope that you found this article resourceful. If you have questions about Macro’s unique AI capabilities or want free access, contact our team of experts.