Products

Our products

Privacy-first AI products and platforms we've built — for government bodies, leading organizations in Israel, and as our own flagship offerings. Each one engineered with privacy-by-design, government-grade security and full data sovereignty.

  • Privacy by design
  • Government-grade security
  • Data sovereignty
  • Frontier innovation

01 / 06

Client product
Abstract illustration of an AI observability platform

AI Observability & Evaluation Platform

2025

Complete visibility into every LLM request, response and decision — Israel-hosted, organization-controlled.

Overview

We built an end-to-end LLM observability and evaluation platform that lets AI teams monitor, score and improve their systems in real time. The platform traces every request, captures the full decision chain, measures quality through automated evaluations, and surfaces business insight on cost and performance.

What we built

Full integration including conversation tracing, versioned prompt management, LLM-as-judge automated evaluations, per-model and per-provider cost and latency monitoring, smart alerts on quality drift, and a developer-grade exploration UI.

Privacy, security & sovereignty

The platform runs exclusively on Israeli infrastructure. All data — prompts, completions, metadata — stays within sovereign boundaries. Per-project and per-team permission isolation, full encryption, and persistent audit logs. Air-gapped deployment is supported for sensitive environments.

Key technologies

LLM TracingEvaluationsPrompt MgmtCost & LatencySelf-Hosted

Innovation & outcomes

This is currently one of the first sovereign observability stacks in Israel — letting organizations adopt AI at scale without losing control, and meeting strict regulatory and privacy requirements.

Outcomes

  • Full visibility into cost and quality across every model in the organization
  • Real-time detection of quality drift — before it reaches the customer
  • Full compliance with Israeli privacy and regulatory requirements
Privacy by designGovernment-grade securityData sovereigntyFrontier innovation
Schedule a consultation

30-minute discovery — explore how this could be built for your organization.

02 / 06

Client product
Abstract illustration of an intelligent model-routing gateway

LLM Gateway & Access Control

2025

One intermediary layer for every AI model — smart routing, API keys, budgets, role-based access and audit trails.

Overview

We built a unified gateway across all providers and models — OpenAI, Anthropic, Google, local models and any future provider — that lets an organization manage access, cost and policy from one place. Instead of dozens of scattered keys, there is one organizational key and granular control.

What we built

Smart task-and-cost-aware routing between models, team- or user-scoped API key issuance, daily and monthly budgets, role-based access control over model groups (chat-pro, coding-fast, embed and so on), full per-request audit logs, and advanced rate limiting. Fully OpenAI-API compatible — every existing client works out of the box.

Privacy, security & sovereignty

All upstream provider keys are stored encrypted on the server and are never exposed to end users. Every request is authenticated, authorized and budget-checked before reaching the upstream provider. Full traffic encryption, persistent records of who used what and when, and instant ability to revoke a user or team.

Key technologies

Model RoutingAPI KeysBudgetsRBACAudit Logs

Innovation & outcomes

An architectural shift: instead of each application holding its own keys and tracking its own costs, the organization manages all of its AI centrally. Result: budget transparency, dramatic cost reduction, and a new-model rollout time measured in seconds rather than days.

Outcomes

  • One organizational API key instead of dozens scattered around
  • Significant reduction in AI cost via smart routing and budgets
  • Roll out a new model org-wide in seconds rather than days
Government-grade securityData sovereigntyFrontier innovation
Schedule a consultation

30-minute discovery — explore how this could be built for your organization.

03 / 06

Client product
Abstract illustration of a private, secured knowledge vault

Sovereign Knowledge Intelligence · Private RAG

2025

The models know what you know — without a single bit ever leaving your organization.

Overview

A Retrieval-Augmented Generation platform hosted entirely on the customer's infrastructure, letting an organization turn its internal knowledge — contracts, case law, protocols, policy documents — into a high-quality conversational search corpus, without sharing a single document with any external provider.

What we built

A full pipeline for ingestion, embedding, vectorization and semantic + lexical search, Hebrew- and Arabic-tuned, with hybrid retrieval combining BM25 and learned dense compression, a precise citation mechanism, document versioning and priority management, and a structured reasoning layer to prevent hallucinations. All in a modular architecture that installs on-premises or in a sovereign cloud.

Privacy, security & sovereignty

The system supports fully air-gapped installation. Documents and embeddings are encrypted, access is controlled at the document and user level, and every query and answer is fully audit-logged. Zero outbound traffic to external APIs while processing sensitive documents.

Key technologies

Sovereign RAGVector SearchHebrew & ArabicAudit TrailAir-gapped

Innovation & outcomes

The combination of Hebrew-tuned hybrid retrieval of unusually high quality, a sovereign architecture, and installable-anywhere deployment makes this a credible Israeli replacement for Western RAG offerings — with no compromise on answer quality.

Outcomes

  • Internal documents become conversational — without a byte leaving the org
  • Semantic-search quality in Hebrew on par with global solutions
  • Air-gapped deployment for the most sensitive defense environments
Privacy by designData sovereigntyGovernment-grade securityFrontier innovation
Schedule a consultation

30-minute discovery — explore how this could be built for your organization.

04 / 06

Client product
Abstract illustration of an accessible bilingual digital assistant

Public-Service AI Assistant

2024–2025

Substantive, accurate and accessible answers for every citizen — in Hebrew and Arabic, around the clock.

Overview

A citizen-facing conversational system for a government body that lets the public get accurate, real-time answers grounded exclusively in the official policy and regulation documents. Every answer is cited, every source is exposed to the citizen, and every interaction is built to WCAG accessibility standards.

What we built

Bilingual (Hebrew + Arabic) conversation UI, focused RAG retrieval over the official document corpus, guardrails that produce reasoned refusals when no reliable source exists, escalation paths to a human agent when needed, and full WCAG 2.1 AA accessibility including screen-reader support, keyboard navigation and high contrast.

Privacy, security & sovereignty

No personally identifiable information is stored without explicit consent. All conversations are logged for system improvement — anonymously. Citizen traffic is encrypted end to end. Compliant with Israeli privacy regulation and accessibility law.

Key technologies

RAG GroundedGuardrailsWCAG 2.1 AAHebrew + ArabicCitizen UX

Innovation & outcomes

Combining quality RAG, source-cited answers, full accessibility and bilingual citizen service sets a new standard for what trustworthy public-sector AI looks like. A citizen who receives an answer also sees where it came from, and why.

Outcomes

  • Quality answers for tens of thousands of citizens daily without a human in the loop
  • Every answer cited and grounded in an official source only
  • Full accessibility — service for every citizen, without exception
Privacy by designFrontier innovationGovernment-grade security
Schedule a consultation

30-minute discovery — explore how this could be built for your organization.

05 / 06

Lawcal product
Abstract illustration of raw documents being transformed into structured evidence

Lawcal Matter

2025

Turns any legal file — scanned, photographed or recorded — into clean, citable, AI-ready evidence.

Overview

Lawcal Matter is the pre-processor that makes your AI tools actually work on real legal files. Drop a scanned 400-page Hebrew contract. Drop a 4-hour court recording. Get back clean, citable output you can paste into ChatGPT or Claude — and have it actually work.

What we built

An OCR engine tuned for Hebrew, Arabic and low-quality scans, integration with Azure Document Intelligence and GPT-5 for cleanup and structure, multi-hour audio transcription with speaker diarization, export to clean formats (Markdown, DOCX, JSON), structured retention of citations and references, and a deliberate legal workflow.

Privacy, security & sovereignty

All files are stored in GCP me-west1 (Israel region) — not a single byte leaves Israeli sovereignty. Built around Israeli attorney–client privilege rules, full per-client isolation, encryption at every stage, and instant full-erasure capability. Validated for legal use by leading Israeli law firms.

Key technologies

Hebrew OCRArabic OCRAudio DiarizationGCP me-west1AI-Ready Output

Innovation & outcomes

This is the missing step between a real legal document and any AI model. Without Matter — the model gets garbage. With Matter — it gets evidence. We are not competing with ChatGPT; we are making it usable in the Israeli legal world.

Outcomes

  • Thousands-of-pages client documents become AI-ready in minutes
  • Hours-long court recordings become citable evidence material
  • All files stay under Israeli sovereignty — privilege-compliant
Privacy by designData sovereigntyGovernment-grade securityFrontier innovation
Schedule a consultation

30-minute discovery — explore how this could be built for your organization.

06 / 06

Lawcal product
Abstract illustration of a unified AI-powered operations control room

Lawcal OS

2025

The unified AI operations platform for your organization — chat, transcription, OCR, agents, all under one brand.

Overview

Lawcal OS is a white-label AI operations platform your organization deploys in 10 minutes. You get a unified, branded, fully RTL-Hebrew interface across every day-to-day AI tool — streaming chat, automated committee-protocol generation from a recording, audio transcription, OCR, a PDF toolkit, and a studio for building custom AI agents.

What we built

Streaming chat UI with conversation management (search, pin, archive, folders), automatic committee-protocol generator with speaker attribution and decisions, bilingual audio transcription, document OCR, a PDF toolkit (merge, split, convert, optimize), a visual studio for building and sharing AI agents, built-in authentication, and full Hebrew + English support.

Privacy, security & sovereignty

Deploys fully inside the organization's own infrastructure or in a sovereign cloud of choice. Authentication via OpenWebUI with enterprise SSO support, full encryption, tenant isolation, and fine-grained permission management. No vendor lock-in — your data stays with you.

Key technologies

Streaming ChatCustom AgentsRTL FirstTranscriptionPDF Toolkit

Innovation & outcomes

Instead of 50 employees each signing up for their own ChatGPT account and leaking private data into foreign clouds — the organization gets one branded, sovereign platform that covers the entire common AI surface. A ready-to-deploy answer for ministries and privacy-sensitive enterprises.

Outcomes

  • A full enterprise AI platform deployed in 10 minutes
  • Instead of 50 private ChatGPT accounts — one branded, secure AI experience
  • A studio for building agents tailored to the organization's unique workflows
Data sovereigntyFrontier innovationPrivacy by designGovernment-grade security
Schedule a consultation

30-minute discovery — explore how this could be built for your organization.

Interested in a similar product?

Whether you're a government body or a private enterprise — let's talk about how a tailored AI product could solve your specific challenge.

Schedule a consultation

Explore More

Our Solutions

AI services for government and legal sectors

About Lawcal AI

Our team, vision, and clients

Contact Us

Book an AI consultation