Production AI. Online or in your perimeter.

Wavenetic builds AI products thousands of users hit on the public web every day — finance, document intelligence, marketing, dev tooling. The same code deploys inside your firewall, on Microsoft Foundry, or on a sealed appliance.

Try our AI Talk to enterprise
EU-built & supported Slovenia (EU)
FREE · PUBLIC · IN SLOVENIAN

Try Slovenia's first public AI CCO.

Ask any question about Slovenian accounting, tax law, or corporate finance. Free. No signup. Live now.

Open the CCO demo Powered by WaveFlow
THE STACK

Production AI. Deployed anywhere.

Wavenetic builds AI products that run inside your perimeter — air-gap, on-prem, private cloud, or Microsoft Foundry. The four below are the architecture choices that make it work everywhere.

Inside your perimeter

Wherever you draw the line, the AI runs inside it. Your servers, your DMZ, a sealed appliance — never on someone else's cloud.

  • No phone-home
  • No outbound telemetry
  • No CDN model pulls

Compliance, handled

AI Act, NIS2, GDPR, DORA — documentation and architecture that meets the bar, shipped with every deployment. Your compliance team gets what they need; you don't get slowed down.

  • Audit logs by default
  • Human-oversight gates
  • Architecture docs out of the box
TRY OUR AI

Real products. Real users. Try them now.

Wave-built AI products in production today. Use them on the open web — or deploy the same code inside your firewall.

WaveFlow
LIVE
AI accountant

WaveFlow

AI-powered accounting for businesses, with Slovenia's first public AI CCO demo for accounting and legal questions.

Try free CCO demo →
cloud · local
LIVE
Document intelligence · yours

WaveOps

Notebooks for your documents. Chat with page-level citations, executive briefings, study guides, mind maps, two-host audio overviews. Claude Sonnet 4.6 / GPT-5 / Gemini 2.5 in cloud mode — or flip to local Ollama with Gemma. Cloud or air-gap on a switch.

Try WaveOps free Sign in →
LAUNCHING MAY 15
AI marketing engine

WaveStorm

Content creation and micro-link AI agency in one. Multi-channel marketing on autopilot.

Get early access
C cl-frontend

✳ Refactoring auth...

C cl-backend

✳ JWT refresh...

X codex-api

6 files changed

A aider-tests

✳ Running tests...

OPEN SOURCE
Multi-agent dev orchestrator

WaveCode

Run Claude Code, Codex, Aider in parallel on your hardware. Apache 2.0 on GitHub.

GitHub → Read more
THE PLATFORM SPINE

One identity. One credit balance. Every WaveAgent.

WaveID and WaveCredits launch May 15 — the universal access layer for every Wavenetic product, public or private.

LAUNCHING MAY 15

WaveID

Single sign-on across every Wavenetic product. Built once, trusted across the portfolio. One identity for WaveOps, WaveFlow, and WaveStorm — public or inside your perimeter.

LAUNCHING MAY 15

WaveCredits

Buy credits once. Spend them on any WaveAgent — finance, documents, marketing, dev tooling. Pay only for what runs.

The same platform our customers use publicly is the same platform we deploy inside your firewall.

Blog

Latest from the blog

View all posts
Browse by topic: WaveFlow On-Premise
Closed-Loop AI Operations: The Control Architecture Nobody Is Shipping
7 May 2026

Closed-Loop AI Operations: The Control Architecture Nobody Is Shipping

Closed-loop AI succeeds or fails on governance, action-layer wiring, and rollback — not model accuracy. Here is the six-stage architecture and the maturity ladder.

Read more
From Public Demo to Air-Gapped Deployment: Building Slovenia's AI CCO
6 May 2026

From Public Demo to Air-Gapped Deployment: Building Slovenia's AI CCO

Slovenia's first public AI CCO for accounting, tax, and compliance is live in WaveFlow as a free public demo — with private cloud, on-premise, and air-gapped deployments available for regulated entities.

Read more
Cloud vs Local AI Agents Is the Wrong Question. Build the Routing Layer.
6 May 2026

Cloud vs Local AI Agents Is the Wrong Question. Build the Routing Layer.

Stop framing AI agents as a cloud-or-local procurement choice. Build a policy-based routing layer that decides per-task where reasoning, memory, tools, and data execute.

Read more
Air-Gapped AI vs. Private AI vs. Confidential AI: What Enterprises Actually Need
6 May 2026

Air-Gapped AI vs. Private AI vs. Confidential AI: What Enterprises Actually Need

Most enterprises asking for air-gapped AI need one of four distinct architectures. Picking the wrong one means paying air-gap prices for cloud-grade risk.

Read more
HOW WE DEPLOY

Same products. Inside your perimeter.

The WaveFlow your accountants try publicly is the WaveFlow we deploy on your ERP. The WaveOps notebooks your team builds on cloud LLMs become air-gap clean the moment you flip the sovereignty switch. Same code. Same engineers. Different perimeter.

Air-gapped appliance

Sealed WaveNode. Zero outbound network. Signed firmware. Offline updates. The default for defence, classified, and critical-infra deployments.

On-premise

Customer servers, customer perimeter, customer control. WaveNode optional. The standard option for regulated enterprise and TSO-class operators.

Private cloud / VPC

Customer-controlled cloud, EU residency, governed access. For organisations with existing cloud commitments and tighter governance than public cloud allows.

Microsoft Foundry

For customers aligned with Microsoft governance. Same Wave product stack, deployed under Microsoft's EU compliance frame. Lowest procurement friction for Microsoft-aligned enterprises.

Same Wave product stack across every option. Same engineers across every deployment. Decision drivers are buyer's compliance posture and procurement preferences — not vendor lock-in.

DOCUMENT INTELLIGENCE

Document intelligence —
yours, not theirs.

Curate sources, chat with citations, and generate briefings, study guides, mind maps and audio overviews — all from one workspace. Flip a switch and the same notebook runs entirely on your own infrastructure.

Try WaveOps free Talk to Wavenetic
CITED ANSWERS
Chat with the sources you choose

Tick the documents that matter, ask anything, get answers that link to the exact page and section.

STUDIO OUTPUTS
Briefings, study guides, audio overviews

One click generates an executive briefing, FAQ, study guide, timeline, mind map, or two-host audio overview.

YOUR MODEL, YOUR CALL
Cloud LLMs or local Ollama

A single switch flips the whole notebook between top-tier cloud models and your on-premise stack.

What ships in WaveOps today

Studio outputs that are the deliverable
Executive briefings, FAQs, study guides, timelines, mind maps, two-host audio overviews. The output is what your reviewer reads — not a chat transcript with the answer buried at message 14.
Page-level citations, jump-to-source
Every claim cites the source page. Click the citation chip — the PDF opens at the right page with the passage highlighted. The flow Google built into NotebookLM, polished for daily work.
Multi-query retrieval + hybrid search
A single user question expands into multiple sub-queries. Hybrid vector + keyword search across all selected sources. Per-source summaries and auto-generated suggested questions seed the next investigation.
BYO-LLM as a feature, not a fork
Cloud LLMs (Claude Sonnet 4.6, GPT-5, Gemini 2.5) for quality. Local Ollama (Gemma + Nomic) for air-gap. Hybrid for the middle ground. Same code path; the switch is in settings.
Enterprise-grade by default
OIDC / SAML federation against your identity provider. Append-only audit log on every query, generation, and source ingest. Self-hostable on a single VM. Your data never leaves your perimeter when the switch is flipped to local.
WAVEOPS IN ACTION
Q3 COMPLIANCE PACK · CITED
QUERY
ANSWER
Page-level citations. Cloud or air-gap on a switch. Append-only audit log.
auto-cycling

Documents in. Answers, briefings, audio out.

WAVE AI PLATFORM

The engine behind every product.

WaveOps, WaveFlow, and WaveStorm are all built on Wave AI — our vertically-integrated platform from models to metal. No wrappers around OpenAI. We built the stack from the ground up.

Domain-tuned models, not general-purpose LLMs
We fine-tune open-weight models on domain-specific corpora: technical manuals, engineering standards, regulatory frameworks. The model already understands your terminology before it ever sees your documents.
Retrieval-augmented generation with citation tracking
Every answer is grounded in your actual documents. Our RAG pipeline retrieves relevant chunks, ranks them by relevance, and generates answers with traceable source references down to page, section, and revision.
Optimised for local GPU inference
Models are quantised and compiled to run efficiently on WaveNode hardware. No internet required. No API latency. Inference happens in-memory, on-site, in milliseconds.
Offline-first update cycle
Model updates, security patches, and new capabilities ship as signed offline packages. Air-gapped deployments stay current without ever opening a network port.
Read our technical deep-dive on the Wavenetic blog
WHO IS IT FOR

Built for organisations that sit on
thousands of documents.

Critical infrastructure operators, public institutions, and universities share the same problem: decades of technical knowledge locked in PDFs, drawings, and legacy archives with no way to search across them.

Defence & intelligence
Air-gapped tactical and analytical AI for classified environments. EU supply chain, no foreign kill-switch surface. Sovereign deployment paths.
Air-gap · Classified · EU-sovereign supply chain
Infrastructure Operators
Critical assets with regulatory compliance, safety procedures, and operational documentation across thousands of sites.
Energy • Transport • Water • Telecom
Regulated enterprise
Banks, insurers, healthcare, pharma manufacturing — for use cases where cloud AI is non-compliant or commercially unacceptable. DORA-ready.
Banking · Insurance · Healthcare · DORA
Sovereign & public sector
Government AI requiring EU supply-chain assurance, full data residency, and conformity-assessment readiness. Procurement-ready.
Government · Public · EU-only supply chain
VERTICALLY INTEGRATED

Hardware to application.
One stack.

Most enterprise AI requires you to stitch together cloud APIs, vector databases, GPU clusters, and custom code. We build the entire stack as one product so it works on day one.

Waves
SOLUTIONS LAYER
Wave Products
WaveOps for operations. WaveFlow for finance. WaveStorm for marketing. Every product ships on the same platform.
Models
MODEL LAYER
Local inference with open models
Optimised for on-site use. No external API calls. Bring your own fine-tunes when needed.
Runtime
RUNTIME LAYER
Wave Runtime
Ingestion, retrieval, RAG pipeline, citations, audit logs, offline updates, and observability. All managed.
Infrastructure
INFRASTRUCTURE LAYER
WaveNode
Purpose-built AI appliances. Edge, rack, and high-density configurations for 24/7 enterprise operation.
One vendor responsible for hardware, software, updates, and support. No integration headaches.
AI INFRASTRUCTURE

WaveNode

We don't just ship hardware. We architect your local AI infrastructure, from a single edge node to a clustered fleet across every site.

WaveNode S
EDGE
WaveNode S

Deploy AI where there's no data centre. Remote substations, field offices, production floors. Runs silently, works offline, fits under a desk.

Fanless, compact form factor
Full offline operation
WaveNode M
RACK
WaveNode M

The workhorse. Rack-mounted, production-grade inference for enterprise teams running WaveOps and other Wave products around the clock.

24/7 enterprise operation
Air-gapped deployment option
WaveNode L
SCALE
WaveNode L

High-density GPU compute for organisations that need multi-model inference, large document corpora, or multi-site clustering under one architecture.

Multi-node clustering
Centralised fleet management
WaveNode cluster mesh topology
Cluster

Nodes cluster. Infrastructure scales.

WaveNodes aren't standalone boxes. They form a managed mesh: replicate knowledge bases across sites, load-balance inference, and roll out updates fleet-wide from a single control plane. We design the topology with you.

Request architecture review Request pricing
SECURITY & COMPLIANCE

Your data stays yours.
Verifiably.

Designed for teams where cloud AI creates more risk than value. Every architecture decision puts data sovereignty first.

Download Security Brief View Architecture Diagram
Optional air-gap
No outbound connections required after deployment. Offline update packages available.
Full audit trail
Every query, every answer, every source citation logged. Export-ready for compliance reviews.
No training on your data
Models run locally. Your documents are never used to train or improve external AI systems.
EU-local support
Installation, training, updates, and escalation handled by our engineering team in Slovenia.
IN PRODUCTION

Receipts, not roadmap.

Live deployments running real workloads today. Customer names withheld pending sign-off — anyone who knows the EU operator landscape will recognise the shape of these references on the second read.

NEXUS at ELES — first AI-native OCPP open platform

Deployed at ELES, Slovenia's national grid operator. Live billing, live charging-session orchestration, used by operations staff every day. The first OCPP control platform built natively around an AI agent runtime.

Substation document copilot

Document intelligence (WaveOps) deployed on-site at a PS-class substation in central Europe. Integrated with operational procedures.

National railway operator

Rolling-stock and depot knowledge platform, in production. Central European TSO/railway-class operator.

Microsoft Foundry deployment

Production traffic for a regulated-enterprise customer. Industry and use case anonymised. Wave product stack on Microsoft governance.

On-site OT integration

Wave product workloads bridged to Modbus, IEC 61850, and OPC UA gateways. The bus end of the stack works because the OT side has done it before.

Naming subject to customer approval. Speak to an engineer for full references under NDA.

PROCUREMENT

Two ways to buy. Same platform.

Designed for enterprise procurement. Predictable costs. Clear accountability.

SUBSCRIPTION
WaveNode as-a-Service

Predictable monthly cost including hardware lifecycle, managed updates, support, and all software. No CapEx.

Hardware replacement included
Managed updates (offline capable)
Priority support SLA
All software included
Best for teams that want zero operational overhead
PURCHASE
Buy + Sovereign Care

Own the appliances outright. Add an annual support contract for updates, security patches, and engineering support.

Offline update packages
Full governance documentation
Asset ownership from day one
Annual support contract optional
Best for organisations with CapEx budgets and internal IT
Request a custom quote
FREQUENTLY ASKED

Questions enterprise buyers ask us first.

Compliance, deployment, security, procurement. The same questions every CISO and procurement officer brings to the call.

Is the AI fully on-premise — or does it phone home?
Fully on-premise. Air-gapped is the default deployment. No outbound network, no telemetry, no CDN model pulls, no license-server check. Updates are delivered through signed offline channels. The customer can verify this on the appliance before deployment and at any point afterwards.
How does Wavenetic handle the EU AI Act?
EU AI Act conformity assessments, technical documentation, audit logs, and human-oversight gates ship with every deployment as part of the product, not a billable add-on. AI used in critical infrastructure, defence, and law enforcement is high-risk under the Act — Wavenetic is built to that bar by default.
Can the same code run on Microsoft Foundry as on a sealed appliance?
Yes. The Wave product stack (WaveOps, WaveFlow, WaveCode, Wave Runtime) deploys across the same code path on Microsoft Foundry, on customer-controlled private cloud, on customer on-premise hardware, and on a sealed WaveNode appliance. The customer chooses the deployment based on their compliance posture; the same engineers ship to all four targets.
What happens to our data?
It stays in your perimeter. No training on customer data. No third-party API calls during inference. No telemetry leaves the network. In on-premise and air-gap deployments, no data is exchanged with Wavenetic at all unless the customer initiates a support session.
Who do we talk to during the deployment?
The engineers who wrote the code. There is no offshore handoff and no escalation tier between the customer and the people who built the product. The named engineers on your contract are the same names you see in the runtime release notes.
How long does an on-premise deployment take?
A typical pilot runs 30 days from approval to live: one process, one customer environment, measurable success criteria. Hardware-included deployments (WaveNode appliances) ship within four weeks of order. Full multi-site enterprise rollouts run 90–180 days from contract.
How are licensing and pricing structured?
Two procurement paths. Subscription: annual licence for the Wave product stack with on-call engineering support, hardware optional. One-time purchase: Wave product licence and WaveNode hardware delivered as a finished appliance, annual support contract optional. Public products (WaveOps, WaveFlow, WaveStorm) move to credit-based billing on May 15 with WaveCredits.
What if we already use Microsoft, AWS, or another cloud?
Then you keep using it. Wave deploys on Microsoft Foundry as a Microsoft partner, and inside customer-controlled VPCs on AWS / Azure / GCP. The on-premise and air-gap options are for workloads where your existing cloud isn't allowed or isn't enough — they're not exclusive of your existing setup.

Have a question we didn't answer? Talk to an engineer.

WHY WE EXIST

We don't rent AI. We build it.

Wavenetic is a European AI product company — not a reseller, not a consultancy billing by the hour. We build our own systems: WaveOps for document intelligence, WaveFlow for finance, WaveCode for engineering teams, and the WaveNode hardware they run on.

The same engineers who write the code install it inside banks, insurers and manufacturers who cannot send a single byte to a public cloud. No outsourcing. No black boxes. No third-party APIs quietly logging your data. You get the box, the source, and the people who built both.

Our products, not reselling
Every line of code, every Wave module, every WaveNode board — built by our team. Full IP, full roadmap control, zero vendor lock-in on our side.
Engineers, not account managers
When you call us, you talk to the person who will write the integration — not a sales rep forwarding tickets to an offshore team.
Your data, your building
Air-gapped by design. Zero egress to public clouds. GDPR and EU AI Act compliance isn't a feature we added — it's the reason we exist.

Built in Ljubljana, Slovenia — shipping across the EU.

ACCEPTING DESIGN PARTNERS

See Wave AI
on your data.

30-minute call. We'll demo our solutions live, walk through the architecture, and size a WaveNode for your deployment.

Or email [email protected] directly.
Wavenetic d.o.o., Ljubljana, Slovenia (EU)