The rush of new start-ups launching to market has not slowed since ChatGPT and GenAI became all the hype. We are lucky at LTH to get the opportunity to meet with and see many of these solutions early, and we’ve been impressed with some of the products that are developing in the ecosystem. What’s particularly exciting is the fact that there are some really thoughtful applications of advanced AI emerging that solve critical problems in different legal practice areas, including pain-points in M&A deal room management and due diligence, law firm matter profiling, enterprise search, litigation management, and drafting.
Here we highlight some of the most promising new start-ups to watch in the GenAI space – these are not your standard foundational LLMs!
The legal industry has been waiting for a new viable player in the enterprise search arena for a long time, and DeepJudge is almost certainly it.
Built by three PHDs in semantic search who have all worked at Google and trained under one of the original Recommind founders, DeepJudge uses a combination of generative AI and conceptual search as well as the UI features that large law firms want to see in enterprise search to develop a platform that delivers better than most of what has launched to market in the past couple of years. DeepJudge sets itself apart by being able to provide scalable contextual knowledge as well as features such as the ability to automatically classify documents according to a taxonomy and data visualizations providing snapshot overviews of document sets.
DeepJudge is deployable on prem or in the cloud, is secure with single sign-on and no data going to third parties, and integrations built out for core document management systems, Sharepoint, HighQ and other relevant platforms. It’s also scalable to hundreds of millions of documents. DeepJudge is a Swiss company but had a presence at ILTACON this year and is making a push into new markets. The DeepJudge knowledge search is already in production with a number of large firms across the US, the UK and Europe. The Generative AI Knowledge Assistant is in beta globally.
Dragon and Lattice
Law firms have been wrestling for years with the problem of matter profiling. Capturing nuanced information about a matter is important for many downstream reasons, including the ability to populate business development databases for pitches and proposals, finding documents from similar matters that can serve as precedents for a current matter, and identifying expertise across a firm. Although firms can work with litigation API providers and companies such as Courtroom Insight to populate some of this data for litigation matters, until now, there has never been a technology solution that has been able to adequately address the matter profiling pain-point for transactional matters. Lawyers are frequently called upon to spend anywhere from one to eight hours populating matter forms at the close of a matter (mostly non-billable work, the completion of which it is notoriously difficult to incentivize lawyers to undertake). A new project by Haley Altman (founder of Doxly, now Litera Transact) and Brett Balmer (founder of Foundation) for Litera is a game-changer, providing the first solution that actually looks likely both to address this problem and to tie together a number of Litera's acquisitions in a way that lends strategic credence to its ever-growing stable. It will also make Litera's Foundation a required platform for most law firms, if it isn't already.
With Dragon (name to be formally announced), legal teams upload final deal documents into the extraction engine which automatically extracts key deal terms relevant to the deal type and allows firms to immediately verify the veracity of the response. Those deal terms are then searchable to find appropriate precedent and identify market trends through an insights interface. Lattice pulls in information from Dragon and pairs it with experience and matter data from Foundation and other firm source systems to allow legal teams to ask complex questions leveraging the firm's collective experience. It is then possible to leverage key information in negotiations: "What break fee terms do we normally get for our clients when acting on the sell-side against [Hall & West] in PE deals in tech?" and when looking for relevant clauses in drafting: “Show me reverse break fees we have negotiated for the sellers in Private Equity deals over $1bn EV in the tech sector which included an earn-out”.
Dragon and Lattice have not yet formally launched. Stay tuned for more information and product demos in November.
A new contract analysis solution, LegalGraph uses a combination of knowledge graph technology and a proprietary large language model to analyze and review lengthy documents to extract key terms and produce reliable summaries, as well as redline contracts.
Unlike some foundational LLMs, LegalGraph is able to process 500+ page contracts, enabling lawyers to rapidly review key obligations to triage, run audits, and close deals.
For example, faced with a lengthy commercial loan agreement, a lawyer can quickly extract clauses and terms such as interest rates, revolving credit loans payment and prepayment conditions, future advance uses and terms, borrower extension options and default clauses, and more – without any additional training required. Similarly, for master service agreements, users can extract terms and clauses such as liabilities and indemnification, intellectual property rights, and any exit strategy terms within minutes. A companion AI legal chatbot works side-by-side with the contract analysis capability, allowing users to query documents for insights as they work.
The solution's knowledge graph technology allows for a more reliable and trustworthy use of AI than a foundational large language model alone (the ex-Meta founders promise that every response is 99.9% correct and grounded on truth using knowledge graphs and their in-house legal LLM, which has been built from scratch). LegalGraph can be deployed either in the cloud or on-premise.
Legau provides knowledge management capabilities and support across all parts of the drafting lifecycle for both litigation and transactional lawyers.
A full knowledge curation workflow combined with legal research properties and generative AI-driven drafting support makes this solution an extremely powerful one for lawyers who seek to leverage the benefits of cutting-edge technology in the place where they already work. Unlike most smart drafting solutions, which typically benefit either transactional attorneys or litigation lawyers, Legau supports both, enabling legal research from third party providers as well as access to internal knowledge bases, and providing clause libraries and recommendations for transactional lawyers.
Users access Legau from a Word plug-in that operates as a side-panel to the drafting window and enables lawyers to undertake their core research tasks, generate outlines and summaries, access precedent documents, build clause libraries and access clause recommendations while drafting.
Legau, a Portuguese company, is making a push into the US this year and may well disrupt the settled landscape around drafting that Litera has established with its Desktop product.
Maxime is a privately trained, privately deployed, secure AI-copilot that empowers legal teams to significantly enhance their capabilities by leveraging their collective knowledge.
Unlike other LLMs on the market, Maxime is a built-from-scratch legal large language model, which means it has been exclusively trained on sound, structured legal content that you can trust. Maxime is transparent about the sources for their legal content and the processes they’ve used to produce a credible Legal LLM. Users are also able to leverage GPT-4 through the Maxime portal, so if they’re looking for answers to more generic questions, this secure solution provides those capabilities too.
In addition to the standard chat interface, Maxime provides UI around specific use cases and is able to perform enhanced document search, contract analysis and clause retrieval. It is also able to interpret your legal data to provide reliable, concise summaries.
Firms that are looking at LLMs to deploy internally should certainly add Maxime to the list of providers they evaluate, alongside CoCounsel, Spellbook, and Harvey.
The midpage legal research and drafting platform leverages GPT to help litigators conduct better research and complete first drafts faster. Instead of using AI to generate documents from scratch (which is problematic in the context of complex documents for many reasons, including lack of credibility), midpage allows lawyers to work the way they normally would but brings together everything they need in one place so that work is able to be conducted more cohesively and efficiently.
On the research side, midpage is plugged into the Free Law Project’s Court Listener data base and provides users with the ability to search across primary sources for caselaw and opinions across the United States. Midpage then uses generative AI to allow lawyers to extract more information from each search result, with tools like drop-down summaries, fact comparisons, and customizable prompts. It also makes it easy for lawyers to organize their research into a "notebook" that can then be used for drafting.
On the drafting side, midpage uses generative AI to help lawyers synthesize their own research, drawing from the hand-selected cases and quotations they saved into their midpage notebook with citations automatically produced in perfect blue-book format. With midpage, the litigator is in the loop every single step of the way – it’s AI-assisted research and drafting that lets you stay in control and verify as you go. Midpage also offers a browser extension and a ChatGPT plug-in.
Those of you who have been following our articles on advanced AI will have read my piece about UI / UX and how important this is in successfully solving particular use cases.
Standd is a company to watch because they are using a combination of large language models (including open-source models and Claude 2) to power a UX-driven platform that has been in development for several years. In other words, the thoughtful interface and workflow predate the release of ChatGPT and are not driven by hype but rather by the subject matter expertise and pragmatic approach of CEO Julie Saltman.
Standd is a modern due diligence product, providing a collaborative data room that complies with industry standards and leverages advanced AI to classify, rename, and maintain ongoing organization of files, as well as providing due diligence review functionality rivalling that of established players in the market. Standd is able to highlight red flags, instantly generate chronologies, timelines, and summaries, and provide lawyers with the ability to rapidly analyze information, surface insights, and gain clarity.
Ideal to support M&A transactions of any size, Standd can ingest large volumes of data and complies with relevant security standards.
What’s Next?
There are more start-ups emerging every day as this tech environment allows for rapid iteration and development. LTH will continue to vet and write about those that look most promising.
We will also be releasing product briefings with each of the solutions covered in this article, to help buyers track developments and identify vendors they may wish to investigate further. Stay tuned!
