USE CASES
A day's legal work,inside Claude or ChatGPT.
No new platform, no uploads. You open the assistant you already use — Claude, Cowork, ChatGPT — and ask. Here are three pieces of real legal work, told the way they actually run.
THREE WALKTHROUGHS
NDA Review
What you say
“Here's the NDA from the counterparty — anything I should worry about before we sign?”
What fires
What happens
You point your assistant at the NDA file. It stays on your machine — only legal-framework queries reach Legalcode. The review runs the structured checklist a transactions lawyer would recognize: party obligations, the definition of confidential information, exclusions, term and survival, governing law.
Each concern comes back annotated with the clause reference and, where relevant, the applicable statute or case law. You accept the analysis, ask for alternative language, or route the flagged items to a lawyer for sign-off — all inside the conversation you started in.
What lands back
- Clause-level risk flags with severity (high / medium / low)
- Jurisdiction-specific enforceability notes
- Suggested redline language for non-standard provisions
Legal Research
What you say
“Can our client terminate for convenience under Icelandic law? Give me the sources.”
What fires
What happens
You ask in plain language. The assistant works out the jurisdiction, the areas of law, and the source types, then retrieves the current text of the applicable statutes and guidance — not summaries, not training data.
It reads what came back, identifies the governing provisions, and synthesises a research memo. Every proposition is pinned to a specific provision with a verifiable reference, so the output is auditable and ready for lawyer review.
What lands back
- Research memo with the governing rule and its exceptions
- Citation list with provision-level references
- Open questions and areas of legal uncertainty flagged
Breach Notification Triage
What you say
“We had an incident overnight — personal data, customers in three EU countries. What do we file, and by when?”
What fires
What happens
You describe the incident to your assistant; the facts stay on your device. The triage works out which notification regimes apply — data subjects' locations, your establishment, sector-specific rules — then runs each regime's threshold test, GDPR's “unlikely to result in a risk” standard included, and returns a notify / no-notify answer with reasoning.
The clock matters here, so everything lands as a deadline matrix with the governing source and deadline basis for each regime.
What lands back
- Jurisdiction scope map — which regimes apply and why
- Deadline matrix for each regulator and affected group
- Required content checklist per notification type
THE FLAGSHIP
And it scales to real matters.
The same vocabulary that triages an NDA can carry a full matter. Our flagship demonstration is a statutory transposition audit: an Icelandic act checked section by section against the EU directive it implements, every divergence traced through the parliamentary record, classified, and stress-tested against counter-arguments.
That is not a demo query — it is the kind of work that fills weeks. It ran as discover, search, fetch, analyze, and trace, composed, inside the assistant — with every finding pinned to a primary source.