Zaphir
Request Demo

DEPLOYMENT

Sovereign where it matters.On-prem everywhere.

Three deployment models. Same AI, same product. What changes is where contracts, policies and matters live and where inference runs.

Talk to salesSee security

THE 3 MODES

Same product, three different perimeters.

Zaphir is the same application across every configuration. The operating perimeter, inference location and hardware ownership change. Our team maps the mode to GDPR, professional-secrecy and data-sovereignty requirements.

01 · ON-PREM

Zaphir On-Prem

A machine dedicated to you, hosted in EU data centres. No shared cloud: your legal and compliance team works in an isolated environment, managed by us, with custom domain and brand.

  • HostingA machine dedicated to you · EU data centers
  • AIInference in EU data centers, EU residency, zero retention
  • ICPLaw firms, scale-ups and compliance teams

For teams that want to start now without managing infrastructure.

02 · DEDICATED ON-PREM

Zaphir Dedicated On-Prem

Zaphir installed in the customer's private cloud · AWS, Azure or GCP, EU region. The app runs under your IAM policies. We handle deployment and runbooks; keys and data stay in your account.

  • HostingCustomer cloud (AWS / Azure / GCP EU region)
  • AILocal on your cloud GPUs, or via API
  • ICPEnterprise legal, regulated groups, public sector

For organisations that already run a private cloud and want deployment inside their own perimeter.

03 · SOVEREIGN

Zaphir Sovereign

Everything on the customer's physical hardware, inside its network. Local inference, air-gapped operation and no outbound calls. Designed for professional secrecy, judicial data and ultra-sensitive matters.

  • HostingCustomer physical hardware, on-premise
  • AILocal on customer GPUs, air-gappable
  • ICPRegulated entities, justice, ultra-sensitive data

For organisations that cannot let a single bit of a matter or confidential document leave the rack.

Not sure which mode fits? A 30-minute scoping call clarifies requirements, integrations and data perimeter.

HARDWARE

Certified enterprise hardware.

For Sovereign, where the institution wants AI on its own rack behind the firewall, we rely on datacenter-grade enterprise hardware. Two reference architectures, picked by workload and budget.

NVIDIA DGX

SMALL · MEDIUM SOVEREIGN

NVIDIA DGX

NVIDIA integrated system for sovereign AI inference. We use it in Sovereign when the workload is small-to-medium. Desktop or rack form factor depending on size.

  • Fully local inference, zero outbound calls
  • Compact form factor, 1U/2U rack depending on size
  • Turnkey setup: arrives configured
  • Available on the Sovereign track (on the customer rack)

ENTERPRISE SOVEREIGN

HP DL380 + 2× NVIDIA RTX 6000 Ada

HP DL380 enterprise server with two NVIDIA RTX 6000 Ada Generation cards. Our enterprise Sovereign reference supports large-scale teams and document workloads.

  • 2× NVIDIA RTX 6000 Ada (96 GB combined VRAM)
  • Hardware cost ~€40-50K, scales to enterprise desks
  • HP 5-year financing at preferential rates available
  • Hardware maintenance and spare parts backed by HP

The HP DL380 + RTX 6000 Ada option is available with HP 5-year financing at preferential rates: it turns a significant capex into a predictable monthly fee, aligned with the IT budget cycle.

THREE COMPONENTS · ONE TECHNOLOGY

Research, Extract and Mail run on every mode.

Zaphir components share the same agentic engine and the same corpus of contracts, policies, regulation and matters. Across deployment modes, products, sources and legal and compliance workflows stay compatible.

01 · CORE

Zaphir Research

The flagship product: agentic search across large legal and regulatory corpora, with every answer cited to source. The backbone for the other modules.

02 · INTELLIGENCE

Zaphir Mail

Mail intelligence for legal and compliance: classifies, prioritises and summarises requests and attachments with source citations in every deployment mode.

One brand, three components, three deploys. No incompatibility between tiers.

FAQ

Six questions, six answers.

The questions our sales team gets in the first 48 hours of a conversation with banks and funds.

On-Prem is a machine dedicated to you but hosted by us in EU data centers: the infrastructure is ours, we manage it, and you work with your brand and domain · AI inference runs in EU data centers, GDPR-safe and zero retention. Sovereign instead runs on the customer's physical hardware, inside their network · DGX Spark or HP DL380 + RTX 6000: the iron is yours, the AI runs on your on-site GPUs, no egress and MNPI that never leaves the perimeter. On-Prem gives you isolation and zero ops overhead; Sovereign gives full technical sovereignty but requires an ops team on your side.

NVIDIA Partner

NVIDIA TECHNOLOGY PARTNER

Reference architectures and AI deployment expertise on NVIDIA infrastructure across all three modes: from On-Prem (DGX managed by us) to Dedicated On-Prem (customer cloud) to Sovereign (DGX or HP DL380 + RTX 6000 Ada on the customer rack).

Which mode is right for you?

Talk to sales