Lex AI - reasoning agent for Serbian legal research
A domain-specialized reasoning agent operating natively inside Paragraf Lex, the de facto legal information platform in Serbia. In production since 2026, deployed under continuous evaluation.
- APPLIED AI ENGINEERING
AI systems for real business workflows.
From complex documents and specialist research to workflow automation and AI agents, we design and deploy production systems around your data, processes and existing software.
- WHAT WE BUILD
AI systems built around the work that needs to get done.
Most engagements start as one of these four. Many end up combining several.
AI Research & Knowledge Systems
Search and reasoning across regulations, internal knowledge, technical material and specialist archives, with sources that can be verified.
Document & Multimodal Intelligence
Turn large volumes of documents, reports, images and unstructured records into structured, searchable information.
AI Agents & Workflow Automation
AI systems that do not just answer questions, but perform defined actions inside existing business processes and software.
Custom AI Products
From the first workflow and data architecture to production deployment, integrations, monitoring and continuous improvement.
- SELECTED WORK
AI systems already operating in production.
Two products of our own and two engagements for the Art Loss Register. Three are live; one is in phased rollout.
Avenos - the operating platform for appointment businesses
The whole booking operation in one system - team calendars, service catalogue, client records, automated reminders and analytics - with a built-in AI receptionist that handles Instagram and WhatsApp on its own, carrying a first message through to a confirmed appointment.
The Catalog Register
An engagement for the Art Loss Register - structuring the historical auction record for due-diligence search. In phased rollout; going live in 2026.
The Classic Car Register
An engagement for the Art Loss Register - linking visual evidence to provenance claims across decades and auction houses. In production.
- WHERE WE FIT
Where AI becomes worth building.
If one of these describes your situation, there is usually a system worth building.
01
Too much specialist information
Your team spends hours searching regulations, archives, reports or internal knowledge.
02
Too much manual document work
Important information is trapped across PDFs, images, scans, emails and databases.
03
Processes that require judgment
The workflow cannot be solved with simple rules, but still contains repeatable reasoning.
04
AI needs to connect to real systems
The result needs to live inside your existing platform, CRM, database or operational workflow.
- HOW WE WORK
Engineering, not prompting.
Our process treats AI systems as engineering problems. The workflow, the data and the definition of a correct result come before model choice - because models change every year, but the data and the domain don't.
01
Understand the problem
We start with the workflow, the data and the definition of a correct result, not with a model.
02
Prove what is uncertain
The assumption carrying the most risk is tested first, on real data and representative cases.
03
Build and integrate
Architecture chosen from results, then wired into the systems and processes you already run.
04
Operate and measure
Quality, cost and failure modes tracked on real usage, feeding back into the evaluation set.
- WHY PROZONE AI
Built to go beyond the prototype.
The hard part is not the demo. It is everything that has to keep working afterwards.
01
Production experience
Systems designed to operate continuously, not just demonstrate capability.
02
Domain-specific engineering
We build around the rules, language and data of the actual business problem.
03
Enterprise software background
AI engineering backed by experience building complex business systems.
04
Long-term ownership
Deployment, evaluation and improvement continue after the initial release.
- TRUST & DEPLOYMENT POSTURE
Built for enterprise environments.
Private deployment, controlled data access and measurable system quality are part of the architecture from day one.
Your data, your infra
On-prem · Private cloud · Hybrid
Audit-ready
Logging · Evaluation · Regression testing
No data exfiltration
Conservative on what leaves your network
Production accountability
Operational ownership, not one-shot delivery
Have a workflow that AI should be doing better?
Tell us how it works today, where the bottleneck is, and what a successful system would need to do. We will tell you whether AI is actually the right solution.