unlisted preview — shared by invitation
Enterprise software, built like it's this decade

Describe it. Watch it become a working system.

Neonyx turns a plain-language description into a real enterprise application — tables, forms, multi-step workflow, role-based access, audit — while you're still in the room. The model is the specification and the running system at once, so they never drift apart.

One engine. Eight enterprise domains. EU-hosted or on your own iron. A real product, not a slide deck.
1 A BA describes it, in plain words
"We track every device by serial, model and site, with run-hours. When calibration goes overdue, the device should lock until a technician signs it off. Technicians only see the queue for their own site; a supervisor approves the return to service."
Neonyx parses
2 It becomes a model
DEVICE SITE WORK ORDER ⟳ workflow
apply · ~90 sec
3 The client uses it
live · real data
FleetOps operations workspace — a real running Neonyx application Meridian Procurement budget board — a real running Neonyx application
8
enterprise domains on one engine
SAP
the founder replaced a production SAP module at a central bank
90s
a description becomes a working system
1 model
the spec and the running system — nothing to drift
Why projects break — and why this one doesn't

The six-month feedback loop, collapsed into one meeting.

Enterprise projects fail in the gap between what the client says and what gets built. The misunderstanding doesn't surface in the meeting — it surfaces in UAT, where fixing it costs twenty times more. Neonyx closes the gap by making the loop short enough that the misunderstanding surfaces in the room.

01

Describe

A business analyst types what the client needs in plain language — no diagramming tool, no spec template.

plain English
02

See it

The entity and workflow diagrams render live as the BA types. The client watches the model take shape.

live diagrams
03

Apply

One click builds real tables, forms, roles and audit — a working system, not a clickable mockup.

real system
04

Fix it live

The client uses it, finds the part you misunderstood, says so — and you change it in thirty seconds.

in the room

This compresses enterprise delivery the way version control compressed collaboration — not by making each step faster, but by collapsing the round-trip.

The market gap
Enterprise tools have depth and bad UX. Modern tools have good UX and no depth.

Nobody has built the combination at scale. Neonyx has — because the engine underneath was built for enterprise from day one, and the AI layer makes it usable by people who aren't software engineers. This is not a lighter alternative to enterprise software. It is enterprise software, finally built like it's this decade — and the depth is real:

· Multi-stage workflow · Conditional validation · Role-based access · Full audit trail · Real-time · Built for regulated work
One engine, every enterprise domain

These are real screens from one platform.

Eight production-grade applications — maintenance, procurement, contracts, real estate, EU funding, capital programmes, a connected medical-device fleet — each built on the same engine, each in its own design language. Open any one to read its full brief.

The thing that makes it work

The model is the spec. The model is the system. One artifact.

Every other approach separates the specification from the software — and the moment they're separated, they drift. Neonyx doesn't separate them. There is nothing to drift.

// reads as

A specification

Stakeholders read the model as the requirements: entities, rules, workflow, roles, in business language.

// runs as

A working system

The same model is the live application — real tables, forms and workflow people log in and use.

// shows as

Living diagrams

And the same model draws its own ER and workflow diagrams, always in sync because they're the same thing.

Change the model and the specification, the system, and the diagrams all change together — automatically, in sync. This is the part that breaks every other method, and the reason a discovery meeting can end with the client using the thing they just described.

Where Neonyx sits

Upstream of the tools you already know.

Neonyx isn't competing with the platforms below — it sits a step before them, solving the problem they each assume is already solved.

vs · AI app builders (Lovable, Bolt, Replit)

Generated code vs. no code to audit

They generate code — and 45% of AI-generated code ships with a top-10 vulnerability, unimproved across model generations. Neonyx generates a model, not code: a governed runtime executes it, so there is no generated code to audit.

vs · Bubble · OutSystems · Mendix · Power Apps

Build without a developer vs. know what to build

Low-code still needs a finished spec — someone has to know what to drag onto the canvas. Neonyx generates the spec and the system together, in the meeting. It's the step before low-code, not a rival to it.

vs · ChatGPT / generic LLM

Describe a system vs. run one

An LLM can describe a system, or write code that compiles. Neonyx produces a working enterprise system — real tables, forms, role-based workflow, audit. The AI is the front door; the framework is the building.

The AI is the front door. The engine — built for regulated work by an engineer with twenty years of it — is the building. The combination is what makes the model real.

What this changes

Different seat at the table, different transformation.

Role
Before
After
Business analyst
50-page Word specs, months of waiting, defending documents in steering committees.
Models live in the meeting and ships a working system — with a captured loop back to the client.
Client / sponsor
Signs off on mockups, then sees the wrong thing six months later.
Uses working software in the meeting and fixes misunderstandings live.
Developer
Implements ambiguous specs, debates interpretation, rebuilds when the client finally sees it.
Receives a model that already runs — and focuses on integrations and edge cases, not boilerplate.
Buyer
Pays for delivery cycles measured in quarters, with a high failure rate.
Pays for cycles measured in weeks, with misunderstandings caught upstream.
When the compliance officer says "show me"

The trust surface is live — not a roadmap promise.

Audit trail on every changeWho changed what, when — captured automatically with the data write.
Governance trail on every AI applyEach AI change is recorded and reviewable, item by item, before anything lands.
EU-sovereign by designEU vendor; EU hosting or fully on-prem — no US-cloud dependency. EU AI Act transparency documented on a public trust page.
Encrypted at restSensitive data is encrypted in storage; user PII handled deterministically for lookup.
One-click export bundleA whole project exports as a complete bundle for review or escrow.
Cost transparent in creditsEvery AI design turn shows its cost; deterministic work is free. No surprise bill.
Who's behind it

Two decades of enterprise systems that couldn't afford to fail.

NA
András Nagy
Founder · creator & principal architect

Neonyx is built by András Nagy — his second-generation enterprise platform, designed from twenty years of delivering systems that risk officers sign off.

For over twenty years he has designed, built and delivered high-stakes systems where there is no room for error: banking back-office, lending and risk workflows, fraud and complaint handling, an electronic document system for a nuclear power plant, and national public-administration platforms — as lead developer and technical lead.

He has also sat on the other side of the table, quality-assuring the enterprise systems that major vendors delivered into some of the country's largest banks and institutions. Neonyx is that experience distilled — with one proof point that's hard to argue with: he replaced a production SAP module for a central bank's cash-logistics operation, signed off by the bank's risk officers.

· MSc, BME — IT & Engineering · MIT Sloan — AI certificate, 2025 · Certified Management Consultant · IPMA-C Project Manager · Banking · public sector · nuclear · e-gov

We compress the loop from "describe the problem" to "use the working system" — from months to minutes, and what gets built is the real thing.

See your own problem modelled

Bring a system you need. Watch it built in the meeting.

A live-modeling workshop: describe a real system from your world, and use it — running, governed, on your data model — before the meeting ends. Fixed fee, credited against a pilot. If it isn't real by the end, it's free.