Moonlit newsletter: May 2026
Welcome to the May edition of Moonlit in Motion, our monthly newsletter. Each month we share what we are building, shipping and learning at Moonlit. Here is everything that happened.
4
Moonlit Connect is live: the API and MCP for European legal data
On May 26 we opened Moonlit Connect, the access layer for the Moonlit Data Layer. Connect ships in two modes: a Data API for teams building legal AI products and an MCP server for legal teams using Claude or any MCP-compatible client.
The Data Layer covers 30 European jurisdictions, 110+ portals, 1000+ sources, 15M+ documents and 32M+ legal references. GDPR alone has 32,000+ incoming references in the reference graph.
Moonlit Data API for engineering teams. Query the Data Layer through a structured REST interface with consistent schemas across all 30 jurisdictions, version-aware responses and the reference graph exposed as first-class objects. See the API docs.
Moonlit MCP server for legal teams. Plug it into Claude (with broader native integrations rolling out as more LLM clients add MCP support) and query the Data Layer in natural language with citations back to the source. €40 per user per month, flat. See the MCP setup guide.
Both come with a two-week free trial against the full Data Layer, not a stripped sample. ISO 27001 certified, GDPR compliant, EU-hosted. Audit logs ship with every query.
On the Data API:
"Moonlit's API/MCP has changed what we are able to build. The breadth, depth and quality of their European legal data is something we rely on every day. It supports our clients to reason across jurisdictions in ways that were not possible before. Questions get answered the same day and the API/MCP evolves alongside what we are shipping." — Saga
On the MCP server:
"Moonlit through Claude has fundamentally changed how I do that research. I formulate a legal question and Claude works out how to break it down, which sources are relevant and reads the full documents. Not summaries, but the actual text. The citations are always correct. I click through to verify them and they have never been wrong. When you are writing a memo that colleagues rely on, that is the only thing that matters." — Lennard van Mierlo, Legal, Quantoz Payments B.V.
Eudia is the first US platform live on Moonlit Connect
Eudia builds Expert Digital Twins for legal and business teams at large enterprises, accelerating high-volume work like contract review and compliance. Their corpus historically covered the jurisdictions where most of their customers operate: the United States. European matters sat behind the same problem every US platform team encounters when it looks at Europe. Building cross-border legal coverage in-house is a multi-year project that needs a dedicated data team and a deep familiarity with European legal sources most US engineers will not have seen before.
Eudia got that coverage through the Moonlit Data API and went live in days.
"The Moonlit Data API gave us a complete European corpus in days, expanding Eudia's legal source coverage across the Atlantic." — Siddharth Saxena, Product Manager, Eudia
If you are building legal AI for the US market and your customers are starting to ask about Europe, Eudia's path shows what is possible.
Citations are now live in Workspaces and Monitors
Open a cell in a Workspace or Monitor and the reasoning now shows numbered citations next to every claim. Click one and the relevant passage pops up, highlighted inside the document, scrolled to the exact line.
Every answer in a cell is traceable to a specific passage in the source, whether that source is a judgment, a regulation or a memo you added.
Two reasons this matters for legal work:
Verifiability. The model is doing the reading, but you still need to see exactly which sentence of the judgment or which paragraph of the regulation the answer came from. Without that, you cannot rely on the cell when you are building a memo, a comparison table or a monitor.
Speed. No more re-opening the source and searching for the passage yourself. The citation does that step for you.
Legal Mind integration is live
Legal Mind users now have direct access to Moonlit's legal data inside their platform.
"The integration with Moonlit is an important step in the development of our product. With this, we are moving further away from standalone tools and apps for legal work and continuing to build a single platform that helps lawyers and legal professionals work faster, more effectively, and more carefully." — George Gowers, Co-founder, Legal Mind
FutureLaw Tallinn: the conversation has moved past "can it draft?"
We spent two days at FutureLaw Conference in Tallinn with 500 people from law firms, regulators and legal-tech vendors. The opening framing set the tone: move beyond AI hype towards verifiable infrastructure.
Five things we took home:
Agents are only as good as the data and context they run on.
Quality of legal data beats quantity.
Combine structured firm knowledge with high-quality legal data. That is where accurate, defensible legal reasoning actually lives.
Legal AI is going agentic, fast. The conversation has shifted from "can it draft?" to "can it act?"
Verification, explainability and auditability are the core pillars of any agent worth deploying.
This is the thesis we have been building Moonlit around. Good to see the experts agreeing.
Coming up: LegalTechTalk London, June 17-18
We will be at LegalTechTalk on June 17 and 18 at the InterContinental London - The O2. Europe's premier event for legal transformation.
We are moving past the experimental phase of LLMs into a period where the quality and accessibility of underlying data sources define the success of legal technology. If that is your problem, come find us.
Industry read: AI models are choking on junk data
A useful Fortune piece on what happens when AI is trained and grounded on low-quality data: hallucinations, drift and confident wrong answers. The answer is not bigger models. It is cleaner, structured, citable sources at the input layer.
This is the bet Moonlit makes. Normalised structure across jurisdictions. A reference graph that connects every rule to every place it is cited. Version history through every amendment. Three streams of enrichment running in parallel: automated extraction, validated user contributions and in-house expert review. The Data Layer is built for AI that has to be right.
Welcome back Giedrius, full-time
We are thrilled to re-welcome Giedrius Mirklys to Moonlit as a full-time member of our data team.
Giedrius joined us in late 2024 alongside his AI Master's at Radboud University. As he now completes his degree, we are proud to have him with us full-time. Over the past months he has made a meaningful impact on one of our most ambitious projects: building the largest European legal corpus. From daily data engineering tasks to the deeper challenges of working with legal data at scale, Giedrius has brought rigour and reliability to the work.
Welcome to the team, Giedrius. This time full-time.
Win an iPad every month by contributing enrichments
We believe the future of law is built on high-quality data and we want you to help us shape it.
Starting this month, we are giving away an iPad to the contributor who submits the most valid enrichment suggestions each month. Enrichments are things like document nicknames, alternative titles and metadata corrections that make Moonlit's legal data more useful for everyone.
How to participate:
Log into the Moonlit platform
Open any document and look for the suggestion button
Submit meaningful data enrichment suggestions
The person who submits the most enrichments each month walks away with a new iPad. Your expertise, rewarded.
We are hiring: working student, content marketing
Hands-on marketing role with real ownership. You will produce blog posts, LinkedIn, newsletters and short-form social, run campaign ops and keep the Moonlit voice consistent across channels.
Looking for a marketing or comms student with exceptional English copywriting, comfortable in Canva (bonus for Figma and LLMs), based in Amsterdam two to three days a week. No legal background needed.
Send your CV and any work you are proud of to info@moonlit.ai.
Follow us on LinkedIn to get the next edition straight in your inbox.
Article written by
Bas Geelen
