Layout-aware OCR
Extracts text from scanned contracts, acts, annexes and tables while preserving document structure.
Technology
The legal AI that doesn't hallucinate.
Zaphir models parties, contracts, clauses, obligations, controls and evidence. Every answer is grounded in documents with precise citations. No opaque generation and no invented sources.
BIRD'S-EYE VIEW
An operational model of your obligations.
Zaphir connects to source systems, recognises parties, contracts, clauses, obligations, policies, controls and evidence, and returns answers with precise citations.
HOW IT WORKS
Four technical foundations, one promise.
Entity recognition, relation mapping, precise citations, grounded generation. Four pieces working together, so every number and every clause is verifiable at the source.
01 · ENTITIES
Every party, role and authority becomes one unified entity.
The same counterparty may appear under a legal name, short name or identifier. Zaphir links it to contracts, obligations, communications and matters in one operational graph.
02 · RELATIONS
A legal ontology rebuilt from your documents.
A party links to contracts, clauses to obligations, obligations to controls and controls to evidence. A live network of relations, not a flat archive. Queries traverse the network while every step remains verifiable.
03 · CITATION
Every answer reports document, page and paragraph.
No opaque generation. Every clause, obligation and finding is anchored to a specific passage: file name, page and paragraph. If the source does not exist, the claim is not generated.
04 · RAG
Retrieval-Augmented Generation, end-to-end.
When a question arrives, Zaphir runs semantic and full-text search across contracts, policies, regulation and matters. Contextual reranking selects the relevant passages for a final answer with automatic citation extraction.
CAPABILITIES
Four functions that work as one.
These are the pieces behind the legal and compliance operating model. They run as a pipeline, each feeding the next.
Multilingual embedding
Vector representations of the legal corpus across Italian, English, French, German and Spanish. Questions and documents can use different languages.
Contextual reranking
Reorders search results by relevance to the specific context of the analysis question. A dedicated model that weighs nuance, not a static score.
NER + entity linking
Recognises parties, roles, authorities, clauses, obligations, amounts and dates, and links them to the right graph entity.
Built today, designed for the years to come.
Numbers that describe how it works and stays traceable under real document workloads.
Millions
pages of contracts, policies and matters indexable in one operational model
Every
answer cites document, page and paragraph
5
languages handled natively in semantic search
100%
of processing in EU data centers
SECURITY
The non-negotiables.
For legal and compliance, security is not a feature list: it is a set of non-negotiable architectural choices.
Data in the European Union
All processing happens in EU data centers. No extra-EU transfer, no exposure to third-country jurisdictions. On-premise deployment also available.
Zero training on your data
Contracts, policies, matter files and correspondence are never used to train models. Zero retention.
End-to-end encryption
Encryption at rest and in transit across the entire flow. Keys managed under European best practices.
Structural GDPR compliance
Privacy by design, native right to be forgotten, DPA available. Compliance is a property of the system, not an afterthought policy.
Complete audit trail
Every operation is logged: queries, access, exports. Immutable logs, exportable to your SIEM, ready for compliance reviews and regulators.
SSO and access segregation
SSO integration and segregation by role, company and matter. Each professional only sees authorised documents, with ethical walls and information barriers enforced.
Ready for a technical evaluation?
Our team can run an architecture deep-dive and proof of concept on a real legal or compliance corpus.