The agent will frame every sub-question, retrieval pass, and the final brief to argue for the selected party, identifying weaknesses in the opposing position and citing Lovdata statutes, ECHR judgments, and Bufdir guidance.

Engine

Both engines run on AWS Bedrock via Claude. Most of the time is spent on multiple question-answering passes — 6–10 sub-questions each requiring a full retrieval and answer cycle. Haiku is faster and handles most cases well. Sonnet produces a more thorough brief with deeper ECHR precedent analysis and stronger multi-party argumentation.

Corpus slices

Three core legal slices are on by default. Enable ECHR Article 9, Hague Convention, Norwegian Courts, Bufdir guidance, or DBN Resources for more targeted research.

Advanced controls
How many angles the agent generates (each framed to strengthen your case).
Corpus chunks retrieved per sub-question.
Minimum similarity for uploaded-doc chunks to be included.
Top sources kept after dedupe + rerank to feed synthesis.
Keep low for grounded legal briefs.

Drop case files here, or

PDF, DOCX, TXT — up to 5 files — chunked + embedded in memory only, never stored.

Ready

Select who you are representing, describe the dispute, and optionally upload case documents. The agent will argue your side — identifying supporting statutes, ECHR judgments, and weaknesses in the opposing position.