Max Junestrand, CEO and co-founder of Legora, discusses how AI is transforming the legal industry—from early experiments with BERT models to deploying agentic workflows that automate end-to-end legal deliverables like due diligence reports. Legora, a Redpoint portfolio company, has raised over $100 million and serves top global law firms by offering a unified platform for reviewing, drafting, researching, and writing legal work.
The Evolution of AI in Law
When Legora started in 2020, early models like BERT were poor for non-English languages (e.g., Swedish), limiting practical use.
The arrival of GPT-3.5 marked a turning point, enabling reliable text generation and reasoning.
Today, Legora uses agentic workflows: LLMs are given access to tools (e.g., document comparison, search), plan multi-step tasks, execute them, and produce final work products—replacing manual processes like physical data room reviews.
Legal work spans a complexity spectrum—from simple data extraction to drafting complex share purchase agreements—and AI is steadily automating the bottom quartile while moving upward.
Why Law Is Uniquely Suited for AI
The legal industry has historically underinvested in software due to misaligned incentives (e.g., hourly billing discourages efficiency).
Work falls into four categories: reviewing, drafting, writing, and researching—each previously served by separate point solutions.
AI enables wall-to-wall platforms like Legora that handle all these tasks, creating pressure on firms to adopt AI both to upskill teams and meet client expectations (e.g., Fiverr’s CEO publicly cited Legora as part of their legal stack).
Hourly billing creates a prisoner’s dilemma: if one firm becomes more efficient via AI, others must follow or risk being seen as overpriced or outdated.
Strategic Advantages of Starting in the Nordics
Legora began in Sweden—a small, fragmented market—which allowed it to refine its product before scaling globally.
As a “fast second mover,” Legora avoided costly bets on training custom LLMs and instead focused on building a user-friendly application layer using off-the-shelf models.
Starting regionally enabled enterprise readiness early: serving large Nordic firms provided referrals to international partner firms and ensured compliance (e.g., SOC, ISO certifications) before entering the U.S. market.
Founders believe starting in the U.S. might have led to narrower product focus due to competitive pressure; in Sweden, they could aim to serve every lawyer across all practice areas.
Product Philosophy: Balancing Today’s Scaffolding with Tomorrow’s Models
Legora builds features that add value today (e.g., playbooks for negotiation rules, citation tracking) even if future models may render them obsolete.
They avoid building what AI labs will eventually offer natively (e.g., citations via API), preferring to deprecate internal code when providers catch up.
Multi-step workflows are shifting from rigid node-based builders to dynamic LLM-generated plans based on high-level instructions and available tools.
To manage model selection, Legora uses Brain Trust for evals and runs classification algorithms to pick the optimal model per task—balancing performance and cost (e.g., avoiding expensive models like o3 for simple tasks).
Pricing and Cost Management
Legora uses a seat-based pricing model for predictability, but faces volatility from variable LLM costs (e.g., one user racked up $10K in a week).
Future pricing may blend platform fees with usage-based components as model costs evolve.
Model costs aren’t decreasing despite improvements—better models like o3 are more expensive—so Legora optimizes by matching model capability to task difficulty.
Infrastructure and Extensibility
MCP (Model Context Protocol) is a key enabler, letting Legora integrate client-specific tools (CRMs, knowledge bases, email systems) directly into AI workflows.
This turns Legora into a platform that extends beyond its native capabilities, allowing clients to bring their own data and templates securely.
Moats and Competitive Differentiation
Legora’s moat lies not in fine-tuning models (which they avoid) but in being a system of record embedded in lawyers’ daily workflows—especially within Microsoft Word, Outlook, and document management systems.
Unlike niche legal tech tools, Legora aims to be the Figma of law: a collaborative platform used by all legal professionals and increasingly by clients too.
Security, privacy, and compliance are critical: all models undergo legal review to meet data processing agreements required by law firms.
Company Building at AI Speed
Legora grew from 10 to 100 people in one year, with onboarding centralized in Stockholm to preserve culture.
The company is fully in-office, emphasizing momentum and urgency. Recruitment is transparent: this is not a 9-to-5 job.
Early commercial focus set them apart in YC: Max sold to European firms at 1–10 a.m. using a ring light to appear professional, prioritizing revenue over pure product development.
They launched early with a compliant ChatGPT-like interface with better RAG on legal docs—good enough to gain traction and iterate weekly.
Market Expansion and Client Expectations
Entering the U.S. required pausing sales for 4–5 months post-funding to refine the product—a rare, mature move for a Series A company.
Reactivating disengaged lawyers is hard; first impressions matter. Reliability (e.g., chunking, RAG accuracy, scalability) had to be solved before onboarding large firms.
Now a system of record, Legora receives urgent Slack/email alerts when issues arise—reflecting deep integration into client deliverables.
Future of Legal Careers
Law firms are shifting from hiring “overachievers who follow instructions” to seeking entrepreneurial, creative thinkers who can challenge legacy processes.
New associates will act as managers of AI agents from day one, requiring fluency in AI tools during legal education.
Firms that don’t upskill risk falling behind as clients demand AI-augmented efficiency.
Quickfire Insights
MCP: Underhyped for enabling universal app integration, overhyped because it’s not yet production-ready (auth, security gaps remain).
Biggest surprise: Users have wildly divergent expectations—from associates expecting full SPA generation via chat to power users with prompt libraries and workflows.
Pivoted early: Dropped a button-based workflow UI (9 colored buttons for tasks) right after YC in favor of a chat interface, deleting 95% of their codebase.
Excited about: AI-powered CROs (Contract Research Organizations) in pharma due to highly manual, data-rich workflows; voice/audio transcription for depositions and client instructions.
Advice for law students: Focus on AI fluency, creativity, and challenging norms—not just diligence and compliance.
