From: redpointai
The legal industry is undergoing a significant transformation with the advent of AI, moving from traditional, fragmented software solutions to integrated, AI-powered platforms. Max Unistron, CEO and co-founder of Legora, a company at the forefront of applying AI to the legal sector, discusses the unique opportunities and challenges in this space. Legora works with many of the top law firms globally and has raised over $100 million, making it one of the fastest-growing AI applications available [00:00:26].
The State of AI in Law
When Legora began in 2020, AI tools were largely ineffective in the legal domain, especially for non-English languages [00:01:40]. The arrival of GPT-3.5 marked a paradigm shift, enabling the transition from pure experimentation to the implementation of end-to-end work deliverables [00:01:52]. For example, due diligence processes, once requiring physical data room visits or specific tools, now involve simply uploading documents to a platform like Legora to find information and generate reports automatically [00:02:08].
Current AI applications in law utilize agents that access tools, plan, and execute tasks, leading to usable end-to-end deliverables [00:02:31]. The primary leverage comes not just from improved models, but from surrounding frameworks like function calling and tool calling [00:02:53]. The historically fragmented legal software market, with separate tools for translations, document comparisons, searching, and reviewing, is now consolidating as AI bakes these functions together [00:03:05].
The complexity of legal work can be mapped on a scale, from simple data extraction at the bottom to complex tasks like drafting share purchase agreements at the top [00:03:25]. AI has already begun to fully automate the bottom quartile, steadily moving towards more complex tasks [00:03:39]. This shift prompts legal professionals to identify where their specific expertise adds the most value, distinguishing between tasks that require human instruction and those where off-the-shelf LLMs suffice [00:03:50].
Why Law is Uniquely Suited for AI
The legal sector has historically seen limited software development, often incentivizing inefficiency over efficiency [00:04:32]. Traditionally, the most advanced tool was a templating system [00:04:50]. Law firms typically handle one-off or complex projects, while in-house councils frequently deal with repetitive tasks like NDA reviews and MSAs [00:04:55]. Legal work broadly involves reviewing, reading, drafting, writing, or researching [00:05:20].
Earlier software solutions focused on narrow point solutions within these categories [00:05:30]. However, AI’s ability to span the entire stack has led to platforms like Legora, which provide wall-to-wall solutions rather than just point solutions [00:05:36].
Challenges and Opportunities for Law Firms
Law firms face pressure to adopt AI to remain competitive. Clients are increasingly using AI tools internally and expect their legal counsel to do the same [00:06:09]. Firms that do not lean into this new paradigm risk not upskilling their teams and losing clients [00:05:57]. The legal software market is valued at 1 trillion market, indicating significant room for disruption [00:06:43].
A long-standing challenge has been the hourly billing model, which could disincentivize efficiency [00:06:50]. However, external pressures like write-offs and price sensitivity mean that clients are less willing to pay for tasks like contract review, especially for large private equity clients [00:07:26]. This has led some large American firms to outsource such work to maintain profitability [00:07:50]. The competitive landscape acts as a prisoner’s dilemma: if competitors adopt AI, others must follow suit to avoid appearing inefficient [00:07:59].
Legora’s Journey and Strategic Decisions
Legora’s competitive advantages include starting in the Nordics, which allowed them to become a dominant player in a smaller, fragmented market before expanding globally [00:08:58]. This approach allowed them to be more ambitious in building a comprehensive product rather than a narrow point solution, servicing a wide range of legal needs [00:11:00]. Being a “fast second mover” enabled them to observe what worked and what didn’t from earlier players, avoiding the mistake of trying to train their own LLMs and instead focusing on the application layer [00:09:40].
Coming from a non-legal background, Legora remained humble and attentive to client feedback, focusing on the evolving relationship between law firms and their clients [00:10:19]. This focus allowed them to build a platform that serves the entire delivery of high-quality legal work, shifting from an internal-facing tool to one that addresses the client relationship [00:10:36].
During Y Combinator, Legora was very commercially focused, launching as soon as they had a viable product—an early version of a private and compliant ChatGPT with better retrieval-augmented generation (RAG) on specific documents and Swedish legislation [00:27:46]. They made a “contrarian move” by deleting 95% of their source code after YC acceptance to pivot from a button and workflow-based interface to a chat-based one, which offered more flexibility for future features [00:39:41].
Product Strategy: Build vs. Leverage
Legora’s product strategy is guided by the principle: if AI labs (like OpenAI, Anthropic, Google) are going to build a feature and make it available, Legora should not build it [00:14:24]. Instead, they focus on building features that add significant value today, even if those might become obsolete as models improve. An example is “playbooks” for negotiation, which might become unnecessary if models can directly process instructions from a Word or Excel file to generate redlines [00:14:46]. Similarly, multi-step workflows, traditionally built with node-based builders, are becoming capable of being generated by LLMs on the fly, given the right tools [00:15:43].
Prioritization is key given the abundance of high-value features. The challenge is to build a cohesive platform rather than a “Frankenstein monster” of disparate functionalities [00:17:31]. This requires careful planning of the platform’s structure, especially from a data perspective, while also adapting to evolving needs of law firms [00:18:08].
Customer Adoption and Training
Teaching lawyers to use AI tools takes significant effort [00:12:44]. Unlike traditional software rollouts with typical 5-10% adoption rates, Legora is seeing 70-80% adoption rates because lawyers are actively seeking out these tools [00:12:57].
User Expectations
A significant surprise in building AI features has been the wide range of user expectations. Some attorneys are “super savvy,” setting up workflows with templates and prompt libraries, while others expect the AI to “write me an SPA” from a simple query, highlighting the need for extensive training and expectation management [00:37:48].
The approach for product launches has shifted from “ship it as soon as we have something” to working with design partners to ensure the product is refined before a proper launch [00:50:50]. This is because users, especially attorneys, do not return after a poor first experience; “you get one chance” [00:29:10]. Reliability and infrastructure, such as retrieval-augmented generation (RAG) systems and handling thousands of users, are critical from the outset [00:29:45].
Pricing Considerations
Legora primarily uses a seat-based pricing model, which is easy for clients to buy and predict costs for [00:19:23]. However, LLM costs can fluctuate wildly, with one user racking up $10,000 in LLM costs in a week [00:19:32]. Over time, Legora envisions a model with a platform fee plus a usage element [00:19:40].
While early thoughts were that LLM prices would continue to decrease, they are becoming better and more expensive, like GPT-4 [00:20:05]. This necessitates using classification algorithms and model pickers to select the most cost-effective model for a given task [00:20:29].
Infrastructure and Tooling
Legora is a major user of LLMs and focuses on building a flexible, future-proof system architecture [00:20:49]. They use tools like “Brain Trust” to run thousands of evaluations on new models, ensuring they meet security, privacy, and legal compliance (data processing agreements) [00:23:44].
A key development is Multi-Agent Coordination and Planning (MCP), which allows Legora to give LLMs access to outside tools, such as redlining documents [00:20:57]. Even more significantly, clients can provide their own tools, enabling Legora to access client-specific CRMs, knowledge databases, templates, or push notifications to client emails, vastly expanding possibilities [00:21:13].
Future of Law and Required Skills
The legal profession is rapidly evolving, with AI augmenting many processes. Law firms are intensely considering how to upskill their new associates and adapt to the lawyer of the future [00:41:48].
Max Unistron
“You previously might have hired, you know, underconfident overachievers that are, you know, really kind of good at following instructions. They’re very thorough and they sort of do things step by step by step. I think now you’re going to need entrepreneurial, creative people who maybe challenge the existing ways things have been done.” [00:42:18]
In this future, lawyers will need to be managers of AI agents from day one [00:43:05]. It will be crucial for every individual to augment their work with AI and demonstrate how they are doing so [00:43:19].
While much of Legora’s work is text-based, future multimodal use cases, particularly with voice and audio transcripts, are exciting [00:40:55]. For example, uploading and transcribing deposition audio files allows for direct interrogation and analysis as if they were written documents [00:41:10].
Company Culture and Growth Velocity
Legora emphasizes high velocity and rapid building, attracting individuals with a strong sense of urgency [00:32:00]. The company culture is one of “momentum breeds momentum,” where people who love winning and hate losing are recruited [00:34:13]. Max Unistron and his co-founder, Sig, lead by example, working long hours to build for the future [00:32:20]. The company is upfront during recruitment that it is “not a 9 to 5 job” and focuses on building rather than maintaining [00:32:48].
Legora has experienced hyper-growth, expanding from 10 to 100 people in a year [00:34:53]. Their first client was a large enterprise, the biggest law firm in the Nordics, which set a high bar for the work they deliver [00:36:18]. From day one, Legora focused on being enterprise-ready, investing half of its initial angel funding in SOC and ISO certifications [00:36:43]. This contrasts with the typical growth path of serving startups or SMBs before moving to enterprise clients [00:36:36].
The current AI era introduces a new expectation for how quickly a software company can grow [00:35:44]. Companies are not just replacing existing software but creating entirely new categories, similar to the internet and mobile eras [00:35:50]. Whoever builds the fastest, with the highest velocity, and the best product and service, will dominate the market [00:36:09].
General AI Industry Insights
Overhyped vs. Underhyped
Max Unistron believes that Multi-Agent Coordination and Planning (MCP) is both underhyped and overhyped [00:37:10].
- Underhyped: MCP is fundamentally enabling universal applications to leverage many different capabilities [00:37:16].
- Overhyped: Everyone is talking about it, but it hasn’t fully moved into production due to challenges like authentication and other technical requirements [00:37:25].
AI in Pharma
Beyond law, Max is passionate about the potential disruption in Contract Research Organizations (CROs) within the pharma industry [00:40:12]. This sector is incredibly manual, involves vast amounts of data, and despite billions paid to slow-moving consultancies, often has structured workflows [00:40:22]. The first fully AI-powered CRO that can ensure and deliver end-to-end results is poised for significant success [00:40:42].
For more information, visit Legora at leguora.com [00:43:53].