How This 25-Year-Old Built A $675M Legal AI Startup (With No Legal Experience)
Max Junestrand, co-founder and CEO of Legora, shares insights on building a successful vertical AI company for the legal industry. He discusses their product strategy, the technical stack designed for a multi-model future, the go-to-market motion for conservative industries, and the challenges of scaling from 10 to 100 people in 13 months.
The rapid evolution of AI means the line between software and service is blurring. For vertical AI companies, the goal is not just to build a tool, but to become a strategic partner that helps customers navigate massive technological transitions. To win, you must align your incentives with your customers, framing it as a shared journey: “we win if you win.”
The Generative AI Catalyst in Legal Tech
Before the advent of powerful Large Language Models (LLMs), legal tech consisted of a fragmented landscape of point solutions. Early models like BERT were ineffective for legal work, especially in non-English languages, as they struggled to understand the semantic meaning of legal clauses when the language varied across documents.
The release of GPT-3.5 was a watershed moment. It unlocked the ability to work with unstructured text, precedent, and complex legal documents, making it possible to consolidate numerous use cases—from contract drafting and review to legal research—onto a single technology stack. This shift allowed Legora to move from a simple MVP to an enterprise-grade system serving tens of thousands of lawyers.
The Legora AI Workspace: A Product Deep Dive
Legora is built as a comprehensive, AI-powered workspace for lawyers, integrating directly into their existing workflows. The product is composed of two main parts: a web application and a Microsoft Word add-in.
Web Application
The web app has evolved beyond a simple “chat with your documents” interface into a sophisticated agent capable of executing multi-step workflows. For example, a user can instruct the agent to:
- Conduct legal research on a specific topic.
- Synthesize the findings and conform them to the firm’s standard language and formatting.
- Draft a complete memo or report.
A key feature is the Tabular Review, a grid-like interface for due diligence. Users can upload hundreds of documents (like employment agreements) and run multiple queries across all of them simultaneously. The system populates a grid with answers (e.g., “Does this contract contain an IP clause?”) and provides direct citations. The engineering challenge here is not just the prompting, but building a robust system that can run hundreds of thousands of queries in parallel, managing chunking, RAG, and the intricacies of legal documents like definitions and cross-references.
Microsoft Word Add-in
Described as “Cursor for lawyers,” the Word add-in brings generative AI directly into the primary drafting environment for legal professionals. A significant feature is Playbooks, a rules-based system for contract negotiation.
- A legal team defines a playbook with rules for specific clauses (e.g., the standard confidentiality definition in an NDA).
- The playbook includes acceptable fallback positions if the primary language is rejected.
- When a user runs the playbook on a document, it automatically redlines the contract to conform to the rules, suggesting approved language.
This feature democratizes legal expertise, allowing non-lawyers, such as sales representatives, to handle initial NDA negotiations, ensuring compliance and consistency while freeing up the legal team for higher-value work.
Technology: A Model-Agnostic and Scalable Approach
Building a durable vertical AI company requires a flexible, model-agnostic architecture.
- Infrastructure: Legora is primarily built on Azure to align with their customers’ existing infrastructure and security requirements.
- Multi-Model Strategy: The platform is designed to be “hot-swappable,” using models from OpenAI, Anthropic (Claude), Google (Gemini), and Mistral interchangeably. This prevents vendor lock-in and ensures that as any underlying model improves, the entire Legora product gets better.
- Efficiency: A classification layer routes queries to the most appropriate model. Simple queries are handled by smaller, faster models, while more complex reasoning is sent to state-of-the-art models. This optimizes both cost and performance.
Go-to-Market: Cracking Sales in a Conservative Industry
Selling to law firms has historically been challenging due to their skepticism and risk-averse nature. However, the market dynamics have shifted. Clients are now demanding more efficiency, creating pressure on firms to adopt AI to stay competitive. This disruption to the “perfect equilibrium” of legal services has opened the door for new technology.
Legora’s sales strategy is not a traditional bottom-up SaaS motion.
- Top-Down Approach: The sale must go through procurement and IT, involving rigorous security and data privacy checks. The buyers are typically the firm’s partners or a dedicated innovation department.
- Land and Expand: Instead of trying to convince an entire partnership at once, the strategy is to start with a single motivated partner or team. By making that team “rock stars” and showcasing their success, demand is generated internally from other teams who want the same advantage.
- Strategic Partnership: The sales process is framed as a long-term partnership. Given the rapid pace of AI development, firms want a partner who can guide them through the transition, not just a software vendor with a five-year lock-in contract. They are betting on Legora’s rate of change and ability to innovate.
On Hypergrowth and Building the Company
Legora scaled from 10 to 100 employees in 13 months, driven by strong product-market fit, which felt like an “infinite pull from the market.”
- Hiring Ex-Founders: A key hiring strategy has been to recruit former startup founders. They bring an invaluable sense of ownership, agency, and a proactive problem-solving attitude that is crucial for a high-velocity environment.
- Scaling Culture: To maintain a cohesive culture across new international hubs (New York, London), the company sends its best people from the Stockholm headquarters to lead the new offices.
- Founder’s Journey: The biggest personal challenge for the founders was the transition from being individual contributors to delegating and hiring people significantly more experienced than themselves in various domains.