Uncapped with Jack AltmanBuilding an AI-Native Software Company With Legora CEO Max Junestrand | Ep. 44
CHAPTERS
- 0:00 – 0:29
‘Blodsmak’ hook: Legora’s intensity (and a translation gone wrong)
Max opens with the Swedish expression “blodsmak” (tasting blood after working extremely hard) and how an English translation made it sound vampiric. The banter sets the tone for Legora’s high-intensity culture and global ambitions.
- •Meaning of “blodsmak” as an intensity/effort metaphor
- •How translation turned it into a vampire-sounding tagline
- •Early signal of Legora’s cultural identity and work cadence
- •Tease of expanding that culture to the US
- 0:29 – 5:18
Why invest despite competition: clarity on model trajectory + legal market intuition
Jack asks Chetan what convinced him to invest when competitors already existed. Chetan explains his prior legal-software experience and why Max’s clear thesis about foundation models and legal data stood out.
- •Existing competitive landscape in legal AI at seed time
- •Investor intuition from prior legal SaaS exposure
- •Max’s conviction that general models would rapidly improve
- •Belief that legal workflows could be transformed if models advanced
- 5:18 – 6:00
Legora’s origin story: from pre-LLM experiments to the post-3.5 inflection
Max recounts the company’s early formation (including founders before he joined) and the shift from BERT-era limitations to LLM viability. The team pivoted from uncertain “AI + law” exploration into a real product company once model capability changed.
- •Company started in 2020 with four co-founders; Max joined later
- •Early BERT/SweBERT limitations and real-world issues
- •LLM arrival (e.g., 3.5) as the decisive pivot point
- •Decision to run hard toward AI+law before the exact product was known
- 6:00 – 6:41
Learning the customer: cold LinkedIn lunches and embedding in a Stockholm law firm
The conversation highlights how Legora built customer understanding without early in-house legal experts. Max describes paying lawyers’ time to learn and working embedded in a firm to understand data and workflows.
- •Cold outreach to lawyers to learn practice-specific problems
- •Engineers-first founding team; first lawyer joined nine months in
- •Embedded, hands-on discovery in a law firm environment
- •Deep understanding of law firm data models and document realities
- 6:41 – 9:05
Why legal adopted AI faster than expected: equilibrium shifts and underserved software
Max explains why law firms moved quickly once one leader adopted AI: competitive pressure forces peers to follow. The group also argues legal was historically underserved by good software, making LLM value immediately obvious.
- •Law firms’ low differentiation creates fast “follow-the-leader” adoption
- •Public adoption by one major firm forces others to respond
- •In-house legal adoption differs from law-firm dynamics
- •Legal workflows had many pent-up problems that LLMs unlocked
- 9:05 – 10:52
Being better than ChatGPT: RAG, citations, guardrails—and moving to an enterprise platform
Max breaks down what made Legora worth buying versus raw foundation models and Copilot. The product evolved from solving basic LLM reliability and workflow issues into a broader platform for high-stakes enterprise legal work.
- •Early differentiation: guardrails, citations, RAG, context handling
- •Operational issues: rate limits, model routing, and task-specific setups
- •Shift from “chat” to enterprise platform transacting large-scale legal work
- •Strategy: give models the right tools/environment + trusted UX for humans
- 10:52 – 14:11
How an AI-native org ships: engineer-led teams, minimal PMs, ‘mini-companies’
Chetan and Max discuss why Legora’s organization looks different from traditional SaaS. Engineering and research lead product discovery, and small autonomous teams build major products like Tabular Review for due diligence at scale.
- •Few traditional product managers; strong engineer/research leadership
- •Need to constantly track model capability changes
- •Tabular Review: parallelized large-document review beyond context limits
- •High concentration of ex-YC founders enabling autonomous execution
- 14:11 – 18:16
The September ‘sprint’ to GA: focus on three use cases and kill the darlings
They describe the pivotal moment before GA where the team narrowed from many potential features to three priority use cases. The result was faster shipping, clearer positioning, and rapid revenue acceleration.
- •Pre-GA moment: too many possible features created diffusion
- •Company-wide alignment session in Sweden to choose focus
- •Doubling down on the few paradigms working in-market (e.g., extraction + Word/Outlook)
- •Outcome: strong revenue growth and readiness for US launch
- 18:16 – 25:11
No sacred cows: roadmap-less building, deleting work, and ‘agents as users’
Max explains Legora’s short planning cycles and willingness to delete large swaths of software as models improve. The team increasingly treats agents as first-class users and designs features for both humans and AI agents.
- •Roadmap horizon shrank from weekly to near-daily reprioritization
- •Cultural requirement: low ego and comfort deleting months of work
- •Models are “no longer the bottleneck”; surrounding software and trust are
- •MCP/agent shift: features must serve human users and agent users
- 25:11 – 27:36
Evals as a moat: customer-contributed tasks, accuracy targets, and ‘once conquered, done’
Legora built deep eval infrastructure to measure model performance on real legal tasks and uncover latent capabilities. Max describes customers supplying real workflows and adopting once accuracy hits required thresholds.
- •Building proprietary eval pipelines early as a competitive advantage
- •Customers contribute real tasks and require specific accuracy milestones
- •Example: LPA key-term report moving from ~60% to 100% accuracy in months
- •Philosophy: if AI can do a task, it will—then the org moves on to harder tasks
- 27:36 – 34:07
Winning pilots and expanding globally in parallel: FDLEs, stickiness, and migrations
The discussion turns to go-to-market: Legora ran competitive pilots that created immediate, visible value and avoided friction in procurement. Forward-deployed legal engineers (FDLEs) drove adoption, and a migration team helped rip-and-replace incumbents.
- •Flexible pilot lengths (30/60/90 days) and strong value delivery
- •FDLE model: tech-savvy lawyers embedded to implement workflows
- •Stickiness driven by usage/workflow cadence more than data lock-in
- •Global expansion without the usual regional sequencing (e.g., India trip)
- 34:07 – 39:20
Starting in Europe as an advantage: multilingual by default and US readiness criteria
They argue Europe forces companies to be multinational from day one, which becomes a structural advantage later. Max shares a practical rule: serve top US firms from Stockholm first, then open a US office.
- •Europe forces multi-country, multi-rule, multi-language product design
- •Benchmark’s view: services markets adopt slowly then rapidly once unlocked
- •Early demos included multilingual + multi-framework support with a tiny team
- •US entry strategy: prove capability with major firms before opening offices
- 39:20 – 47:14
Stockholm culture at scale: onboarding in HQ, consistent office DNA, and ‘#blodsmak’
They describe what makes the Legora office feel different: engagement, humility, speed, and intensity. The company preserves culture by requiring interviewing and onboarding in Stockholm and seeding new offices with culture carriers.
- •High buy-in, cadence, and humility required for AI-native work
- •Centralized onboarding in Stockholm for every employee globally
- •Seeding New York/London (and beyond) with Stockholm culture carriers
- •‘Blodsmak’ becomes a shared meme for intensity and winning mindset
- 47:14 – 49:58
Series D and fundraising philosophy: preempts, dilution discipline, and new CFO energy
Max closes by discussing Legora’s fundraising approach—often preempted rounds and even taking lower term sheets—plus the recent Series D process. He highlights partnering with a new CFO and the oversubscribed demand for the round.
- •History of preempted rounds and limited outbound fundraising
- •Strong stance on dilution and ownership negotiations
- •Series D: first full process with a deck; heavily oversubscribed
- •New CFO from Vanta and the operational ramp for the next phase
