Skip to main content

Private AI for
European enterprise.

Your models. Your data. Your infrastructure. We build AI that runs on your terms — on-premise or EU cloud, GDPR-native, from Trieste.

-89%
Classification error
reduction with KOI
180%
ROI for
enterprise clients
100%
Data stays
under your control
Trusted by
Samsung, Università di Torino, T&B Associati, Podartis
Private infrastructure
PRISMA — Private Intelligence Stack for Modular AI

The stack that makes private AI possible

PRISMA is our private infrastructure for running AI models on-premise or in EU cloud. No data leaves your perimeter. It powers every product we ship. Co-funded by the FVG Region through PR-FESR grants.

On-Premise or EU Cloud

Runs on your servers or on European cloud infrastructure. Your data never leaves your perimeter.

GDPR · AI ACT

Private Models

LLMs selected and fine-tuned for your use case. No data shared with third parties. Ever.

E2E ENCRYPTION

Modular Architecture

Every HTX product runs on PRISMA and adds domain-specific logic on top. Swap, extend, or customize.

PR-FESR FVG
ISO 9001 — in progress ISO 13485 — in progress ISO 27001 — in progress IEC 62304 — medical device software
Real-world deployments

Results. Not prototypes.

Podartis
800 pages of PDF orders into ERP — automatically

End-to-end automation: PDF orders imported directly into the client's new ERP system, with zero manual data entry.

800 PDF pages processed
Direct ERP import
Anomalies flagged for human review
T&B Associati
Excel to ERP: 50 person-days of work done in 1.5

Automated migration from Excel to management software — thousands of cells filled intelligently.

52,000 hours entered
23,000 cells completed
1.5 days vs 50 person-days · 180% ROI
Arethusa
Automatic candidate import into the HR platform

Automated pipeline that accesses shared company mailboxes and enables dynamic mapping with the HR management system.

400+ candidates imported and counting
PDF check for essential information
Intelligent matching control

Sound familiar?
Let's talk.

Tell us what you're working on. We'll tell you if we can help — no strings attached.

Get in touch
HUMAN TECHNOLOGY EXCELLENCE

AI that actually helps people

We don't just ship software. We build AI tools that eliminate busywork, catch critical errors, and give people their time back.

Clinical safety

KOI cuts pre-anesthesia classification errors by 89% — giving anesthesiologists AI-powered decision support on the ASA-PS scale.

Safer drug preparation

The ChemoMaker robot has been integrated with MANTA to enable dialogue and identify critical issues.

Oncology nutrition

ESPEN guidelines, patient data and recipes combined by AI to support clinicians in drafting oncology nutrition plans.

Made in Italy

Built in Trieste — Europe's research capital

Trieste has the highest concentration of researchers in Europe — 4x the EU average. A unique ecosystem of international research institutes at the crossroads of Italy, Central Europe, and the Balkans.

HTX was born here: a 100% Italian AI company building solutions where technology works for people, not the other way around.

The city is at the center of a new innovation wave — from world-class public research to private initiatives like the Agorai Innovation Hub. Trieste is becoming a European hub for AI that improves quality of life. That's exactly what we do.

4x
Researcher density vs EU average
37+
Research institutes in Trieste
100%
Italian, European, independent
GDPR
AI Act compliant by design
Featured project

AI Ecosystem for Clinical Nutrition

Academic research meets culinary excellence meets AI — built to improve how patients eat and recover.

University of Turin Samsung Italia Chef Cracco
01

ESPEN guidelines via AI agents

Clinical nutrition guidelines digitized and queryable in natural language.

02

Smart recipe management

Co-developed with Chef Cracco.

03

AI-optimized meal plans

Generates personalized nutrition plans from patient data, cross-referencing recipes, foods, and validated clinical databases.

HOW WE WORK

Three steps to production AI

01

Analyze

We study your process, your data, and your constraints. We find where AI creates real value.

30 days
02

Pilot

We build a working prototype on your actual use case. Measurable results — not slide decks.

2-4 weeks
03

Scale

Production deployment, system integration, team training. You set the pace.

Your pace
FROM THE BLOG

Why your business needs private AI

77% of employees paste company data into ChatGPT. Italy has already fined OpenAI EUR 15 million. The AI Act imposes new obligations. Find out why private AI is the only safe path.

Read the article

Ready to build
your private AI?

No slide decks. No vague promises. Just a conversation about what AI can do for your business.

Book a call

From our tech blog

Why your business needs private AI (not ChatGPT)
Why your business needs private AI (not ChatGPT)
·1110 words·6 mins
Articoli AI Privacy GDPR Best Practices
Keycloak
Keycloak
·1548 words·8 mins
Articoli API Tech
GitHub - zai-org/GLM-OCR: GLM-OCR: Accurate × Fast × Comprehensive
GitHub - zai-org/GLM-OCR: GLM-OCR: Accurate × Fast × Comprehensive
·1046 words·5 mins
GitHub AI Open Source Python
GitHub - EricLBuehler/mistral.rs: Fast, flexible LLM inference
GitHub - EricLBuehler/mistral.rs: Fast, flexible LLM inference
·1029 words·5 mins
GitHub LLM Rust Open Source
GitHub - alexziskind1/llama-throughput-lab: Interactive launcher and benchmarking framework for llama.cpp server throughput, featuring tests, sweeps, and round-robin load tools.
GitHub - alexziskind1/llama-throughput-lab: Interactive launcher and benchmarking framework for llama.cpp server throughput, featuring tests, sweeps, and round-robin load tools.
·1071 words·6 mins
GitHub Tool Open Source Python