--- title: Building Legal Agents When You Can't See the Data -... description: Jacob Lauritzen, CTO of legal AI company Legora, explains how the team builds and evaluates its agent without ever seeing client data. Models already handle... canonical: https://daily.dev/posts/building-legal-agents-when-you-can-t-see-the-data---jacob-lauritzen-cto-of-legora-0bg2vqfqo twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Building Legal Agents When You Can't See the Data - Jacob Lauritzen, CTO of Legora | daily.dev og:description: Jacob Lauritzen, CTO of legal AI company Legora, explains how the team builds and evaluates its agent without ever seeing client data. Models already handle... og:url: https://daily.dev/posts/building-legal-agents-when-you-can-t-see-the-data---jacob-lauritzen-cto-of-legora-0bg2vqfqo og:image: https://api.daily.dev/og/posts/0BG2vQFqO.png og:image:alt: Building Legal Agents When You Can't See the Data - Jacob Lauritzen, CTO of Legora og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Building Legal Agents When You Can't See the Data - Jacob Lauritzen, CTO of Legora **[Latent Space](https://daily.dev/sources/latentspace)** · 5 min read · 1 upvotes · 0 comments ## Summary Jacob Lauritzen, CTO of legal AI company Legora, explains how the team builds and evaluates its agent without ever seeing client data. Models already handle about 60% of legal work well, he estimates, with the rest needing a smarter model. A citation accuracy problem jumped from 70% to 99% after a single model release, without Legora changing anything. The company runs everything on Azure for data privacy, skips ElevenLabs for transcription, and built Tabular Review, a million-cell spreadsheet feature other legal tech companies copied down to the name. Legora now has 100 engineers, and Jacob still feels short-handed. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How much of legal work can AI models handle well right now? Legora's CTO Jacob Lauritzen estimates that models already perform about 60% of legal work well. The remaining work requires a smarter model to handle reliably. This estimate comes from building and testing legal agents at Legora, a legal tech company that has grown rapidly since starting in a Stockholm basement. _Teams evaluating how far AI can go in specialized domains like law can track real-world progress reports on daily.dev._ ### How can you test an AI agent's performance when you can't see the data it processes? Legora builds test environments modeled on customer work, similar to the simulated 'worlds' used in reinforcement learning, using a team of three engineers and three lawyers. Because client data is confidential, Legora paraphrases traces and narrows failures down to 10 or 20 types rather than inspecting raw outputs directly. _Developers building agents under strict data privacy constraints can follow similar evaluation approaches on daily.dev._ ### Why would an AI agent's citation accuracy suddenly jump from 70% to 99%? At Legora, a citation accuracy problem went from 70% correct to 99% correct after a single underlying model release, without the Legora team changing anything in their own system. This illustrates how much agent reliability can depend on the base model's improvements rather than application-level engineering. _Anyone building on top of foundation models can track how upstream model releases shift downstream reliability on daily.dev._ ## Similar posts on daily.dev - [Building AI for Lawyers - Jacob Lauritzen, CTO of Legora](https://daily.dev/posts/building-ai-for-lawyers---jacob-lauritzen-cto-of-legora-s9tyhxq0d) · Latent Space · 0 upvotes · 0 comments - [Legora is building the agentic operating system for legal work](https://daily.dev/posts/legora-is-building-the-agentic-operating-system-for-legal-work-smkj4zc2x) · OpenAI · 1 upvotes · 0 comments - [Legal AI startup Legora raises $550M Series E at $8.8B valuation](https://daily.dev/posts/legal-ai-startup-legora-raises-550m-series-e-at-8-8b-valuation-pvkgm0qo0) · TechCrunch · 0 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#azure](https://daily.dev/tags/azure), [#rag](https://daily.dev/tags/rag), [#legal](https://daily.dev/tags/legal) [View this post on daily.dev](https://daily.dev/posts/building-legal-agents-when-you-can-t-see-the-data---jacob-lauritzen-cto-of-legora-0bg2vqfqo) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Building Legal Agents When You Can't See the Data - Jacob Lauritzen, CTO of Legora","url":"https://daily.dev/posts/building-legal-agents-when-you-can-t-see-the-data---jacob-lauritzen-cto-of-legora-0bg2vqfqo","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/building-legal-agents-when-you-can-t-see-the-data---jacob-lauritzen-cto-of-legora-0bg2vqfqo"},"datePublished":"2026-11-02T22:58:32.332Z","dateModified":"2026-11-02T22:58:59.747Z","description":"Jacob Lauritzen, CTO of legal AI company Legora, explains how the team builds and evaluates its agent without ever seeing client data. Models already handle...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/f67242efde9e421ecdbfacea4be5f7b4?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/f67242efde9e421ecdbfacea4be5f7b4?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Latent Space","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Latent Space","logo":"https://media.daily.dev/image/upload/s--DnJ9laFj--/f_auto,q_auto/v1780213196/logos/latentspace?_a=BAMAMiWQ0","url":"https://daily.dev/sources/latentspace"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/building-legal-agents-when-you-can-t-see-the-data---jacob-lauritzen-cto-of-legora-0bg2vqfqo","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,azure,rag,legal","timeRequired":"PT5M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Latent Space","item":"https://daily.dev/sources/latentspace"},{"@type":"ListItem","position":3,"name":"Building Legal Agents When You Can't See the Data - Jacob Lauritzen, CTO of Legora"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/building-legal-agents-when-you-can-t-see-the-data---jacob-lauritzen-cto-of-legora-0bg2vqfqo#faq","mainEntity":[{"@type":"Question","name":"How much of legal work can AI models handle well right now?","acceptedAnswer":{"@type":"Answer","text":"Legora's CTO Jacob Lauritzen estimates that models already perform about 60% of legal work well. The remaining work requires a smarter model to handle reliably. This estimate comes from building and testing legal agents at Legora, a legal tech company that has grown rapidly since starting in a Stockholm basement. Teams evaluating how far AI can go in specialized domains like law can track real-world progress reports on daily.dev."}},{"@type":"Question","name":"How can you test an AI agent's performance when you can't see the data it processes?","acceptedAnswer":{"@type":"Answer","text":"Legora builds test environments modeled on customer work, similar to the simulated 'worlds' used in reinforcement learning, using a team of three engineers and three lawyers. Because client data is confidential, Legora paraphrases traces and narrows failures down to 10 or 20 types rather than inspecting raw outputs directly. Developers building agents under strict data privacy constraints can follow similar evaluation approaches on daily.dev."}},{"@type":"Question","name":"Why would an AI agent's citation accuracy suddenly jump from 70% to 99%?","acceptedAnswer":{"@type":"Answer","text":"At Legora, a citation accuracy problem went from 70% correct to 99% correct after a single underlying model release, without the Legora team changing anything in their own system. This illustrates how much agent reliability can depend on the base model's improvements rather than application-level engineering. Anyone building on top of foundation models can track how upstream model releases shift downstream reliability on daily.dev."}}]} ```