42 %
of German companies: staff use private AI outside IT
Bitkom 2025
Company AI from Austria · EU-hosted
Company AI with a filter before sending.
100+ AI models. One company account.
Your team works with the leading AI models in one company account. What runs through HOVIGuard is checked before sending and logged: IBANs, credit cards and API keys are blocked before a model sees them. Hosted exclusively in the EU.
No credit card. Check it yourself in 2 minutes: paste a test IBAN, see the block, open the log.
Filter before sendingNo model trainingEU · Made in Austria
MultiLayer Data Shieldactive
1Privacy Filteractive
2Content Filteractive
3Manipulation Filteractive
4Audit Logactive
Made in Austria
Made for Europe
EU standards built inEU-only hosting · GermanyMultiLayer Data Shield · 4 protection layersData filter before sendingNo training use of your dataNo data in US data centresIBAN, credit card, API keys blockedMade in Austria
Your AI is already running —
just not under your control.
Your team already uses AI — whether you allow it or not. The question isn't if, it's where: in private ChatGPT accounts, with no audit, no filter, exposing your corporate IP.
No controlShadow AI
With HOVIGuardA vetted AI infrastructure
78 %
of AI users bring their own unapproved tools · Bring Your Own AI
Microsoft 2024
up to €20,000,000
GDPR fine for serious violations — or 4% of global annual revenue
Art. 83(5) GDPR
34.8%
of company data entered into AI tools is sensitive — and rising
Cyberhaven 2025
63 %
of breached organizations have no finished AI governance
IBM 2025
Shadow AI? A thing of the past. With HOVIGuard, invisible shadow AI becomes a central, vetted and auditable AI infrastructure — hosted in the EU, with no training on your data.
HOW IT WORKS
Verified before
data flows.
Behind every AI request stands your company — with its IP. HOVIGuard checks every request before it reaches a model.
Block
Mask
Confirm
Individually configurable — you define your own blocklist (terms such as project or client names) and choose per term: block, mask, or ask for confirmation.
Every request checkedEU hostingAudit trail
MultiLayer Data ShieldFour layers between your
Four layers between your
team and the AI.
Every employee request passes through automatic verification layers — before reaching an AI model. Sensitive data detected, problematic content filtered, compliance risks stopped before the model, every action logged for audit.
Layer 01
Privacy Filter
Bank data (IBAN, credit card), ID and insurance numbers, credentials (API keys, passwords) — blocked before reaching the AI model; e-mail addresses masked.
IBANCredit cardID numberAPI keysE-mail+ many more
Layer 02
Content Filter
Insults, discrimination, illegal content, drug references — requests are stopped before the AI or submitted for confirmation.
HateDrugsExtremismDiscrim.Illegal+ many more
Layer 03
Manipulation & Behavior Filter
Self-harm, manipulation attempts ("ignore all instructions…") and harmful requests — blocked or submitted for confirmation per config.
ManipulationEthicsJailbreakSelf-harmPrompt-Inj.+ many more
Layer 04
Audit Trail
Who used which model when and for what — logged and exportable as basis for your own compliance evidence.
CompleteExportableGDPRForensics+ many more
AI model · verified
| # | Principle | Meaning |
|---|---|---|
| 1 | Lawfulness | Processing with a clear legal basis |
| 2 | Purpose limitation | Use data only for defined purposes |
| 3 | Data minimisation | Collect only as much data as necessary |
| 4 | Accuracy | Keep data current and correct |
| 5 | Storage limitation | Don't store longer than needed |
| 6 | Integrity & confidentiality | Protection against loss and manipulation |
FeaturesMore than a chat.
More than a chat.
One platform.
From multi-model chat via audit dashboard to image and video generation — all in one verified, EU-hosted environment.
Feature 01 · Chat
Multi-Model Chat
Switch between Claude, GPT, Gemini, Mistral per request. One interface, all models.
Feature 02 · Veritas — NEW
Veritas Chat — 5 AIs, one consensus
Anti-hallucination protection through multi-model consensus. 5 independent AIs check every answer, a leader delivers only what the majority agrees on — disagreements transparent.
Feature 03 · Audit
Audit Dashboard
Who used which model when and for what — fully traceable, exportable.
Feature 04 · Compare
Model Comparison
Same question to multiple models in parallel — direct output comparison in seconds.
Feature 05 · Image
Image Generation
Google Gemini Image (Nano Banana) in four variants — directly in workflow, EU-hosted, with audit trail. Templates for style, business, entertainment, background — plus a prompt assistant that gets the most out of your idea.
Feature 06 · Video
Video Generation
Google Veo 3.1, Veo 3, Nova Reel — cinema-ready 720p clips with native sound, EU-hosted. Templates for style, ads, entertainment, background — format and tonality are pre-set.
Feature 07 · Pilot
Model Pilot
Automatically picks the best model for every request — code, vision, search, reasoning.
The difference
With or without HOVIGuard.
See at a glance what changes for your business.
✕ Without HOVIGuard
Shadow AI & data leaks
- Employees copy customer data into private ChatGPT accounts
- Inputs often flow into model training (free/consumer plans)
- Corporate IP ends up in third-party provider logs
- Data frequently leaves the EU
- Shadow AI: no overview which employee uses which tool
- No audit trails for privacy requests or regulator audits
- High fine risks on privacy and AI violations
✓ With HOVIGuard
Company AI with a filter
- Inputs do not train AI models — contractually guaranteed
- Data does not leave the EU
- MultiLayer Data Shield: PII, brand, content filters before the model
- IBAN, credit card, API keys blocked, e-mail masked — before the model sees it
- Complete audit trail — exportable any time
- 100+ AI models, continuously updated: Claude, GPT, Gemini, Mistral on one platform
- Model Pilot auto-routes — code, vision, reasoning, image, video
- One central, approved platform for your team
- Reputation protection for corporate IP
- Made in Austria — control for IT & data protection
For every role
Who benefits? Everyone.
From CFO to employee — every role gets what they need.
GF
Executives
One company account for the whole team. Every request is logged — your answer when the data protection officer asks.
IT
IT & Security
Central control as alternative to Shadow AI (uncontrolled use of external AI tools in your team). Per-group policies, granular permissions, audit log of all requests.
CO
Compliance & Legal
Data processing agreement under Art. 28 GDPR, data processing in the EU, every request logged.
MA
Employees
One platform for Claude, GPT, Gemini & Co. — work faster, without tool-switching.
100+ models · EU-hosted100+ leading AI models, continuously updated.
100+ leading AI models, continuously updated.
One platform. One invoice.
Mistral, Claude, GPT, Gemini, DeepSeek, Llama — connected via EU infrastructure, protected by the MultiLayer Data Shield.
Our promise
A company AI that checks before sending and logs every request. What HOVIGuard can do, you can verify yourself in 7 days.
KP
Karl Pilz · Founder
Pricing
Three paths to secure enterprise AI.
From free trial to a custom license. Transparent pricing, no minimum term.
Demo
Free
7
Filter, blocklist and log usable — with 5 trial models.
- 1 member · 7-day duration
- 5 trial AI models
- Filter, blocklist and log usable
- Audit trail included
- EU hosting, no training-data use
- Veritas Chat available (limited trial budget)
Pro
Pro-Plan
€25
Self-service. Full platform for productive teams.
- 1–100 members
- 100+ EU models (chat, image, video, code, audio)
- Model Pilot & Model Comparison
- Complete audit trail & RBAC
- Blocklist for your own terms (project, client names)
- Email support · monthly cancellable
- Veritas Chat: anti-hallucination consensus from 5 AIs
Enterprise
Custom
€on request
Custom volume license and personal support. Further requirements (e.g. SSO) on request.
- Everything in Pro
- More than 100 members · custom volume license
- SSO integration on request
- Special requirements on request
- Individual agreements on support and term
- Personal point of contact
FAQ
Answers to the most common questions.
What is HOVIGuard?
HOVIGuard is a secure AI platform from Austria. Several leading AI models are available in one app. More on the Features overview.
Which AI models are available?
A selection of common models from well-known providers is integrated — kept up to date. A current overview is on the Features page.
Are my inputs used for AI training?
Inputs are not released for model training. Details in the Data Processing Agreement (DPA).
Where is the data processed?
Data is processed within the EU. More in the Security & Compliance section.
How does HOVIGuard protect sensitive inputs?
Multiple protection layers check inputs automatically — e.g. for personal data or problematic content. More on the MultiLayer Data Shield page.
What is an audit trail and what is it for?
An overview logs AI requests with timestamp, model and result — as a basis for internal traceability. More on the Traceability page.
How is the right AI model picked?
You can pick the model yourself or use the auto-selection. Details on the Model Pilot page.
Can models be controlled per team or group?
Such settings can typically be configured individually. Details on the Enterprise solution page.
Does HOVIGuard work with our sign-on (SSO)?
SSO integrations can be discussed — requests via the inquiry page.
Is HOVIGuard available for our own data center?
Custom setups are generally possible. Requests via the inquiry page.
What does HOVIGuard cost for companies?
Current pricing and plan comparison are on the Pricing page. For larger setups we provide custom quotes.
Do I get a Data Processing Agreement (DPA)?
A template is available on the DPA page. We provide the signed agreement on request.
How can I try HOVIGuard?
7 days free, no credit card. Register now. For enterprise configurations with more seats, request a custom quote.
Ready for secure
enterprise AI?
Try free for 7 days, no credit card. All features, all models, complete audit trail.