Cynara Software House. Enterprise Engineering.

Accelerated by AI. Grounded in engineering.

Build faster. Stay in control.

We build and modernize business software at AI speed, with engineers who stand behind every release.

Take the 4-minute AI check

AI can write code.

It can't take responsibility for the system.

We use AI wherever it makes engineering faster. People stay accountable for the architecture, security, reliability, and the decisions that matter.

Four ways we work.

  1. Build

    Ship faster without lowering the bar.

    New products, smarter features in software you already run, or complete systems where AI is one part of the answer. When deterministic code is the better choice, that is what we build.

    • Prototype to production
    • Built to own, not rent
    • Documented for your team

  2. Advise

    Know what to build before you build it.

    A vendor-neutral view of where AI creates value in your business, and where conventional software is the better choice. Build, buy, or don't build at all.

    • Opportunity and readiness reviews
    • Build-vs-buy decisions
    • Never tied to a vendor

    What the 4-minute check gives you

    Way of workingData &complianceQualityContinuityCost &ownershipDirection
    A profile in six areas, from sample answers.See the sample result

  3. Own

    Models on infrastructure you control.

    Open-weight models on your own hardware or inside your own cloud account, from a single deployment to DGX and multi-GPU clusters.

    • On-prem, cloud, or hybrid
    • Predictable operating cost
    • Data stays in your network
    Your data, the model, and your application stay inside your network. Cloud AI APIs stay outside it.

  4. Connect

    Put AI inside the systems you already use.

    Agents wired into your CRM, ERP, databases, and internal tools over open standards like MCP, with permissions and an audit trail.

    • Standards-based (MCP)
    • Works with your stack
    • Portable, no lock-in

Don't start with AI. Start with the problem.

Sometimes the right answer is an AI agent. Sometimes it's an API, a better workflow, or nothing at all. We'll tell you which before you spend anything.

Four stages. Each one ends with something you keep.

At the end of every stage, you decide whether to go further, and everything built so far is already yours.

  1. AI Readiness Review

    Fixed price · 2–3 weeks

    Find out what's worth doing. We look at your processes, data, and systems, and at how your team specifies, builds, tests, and releases software. You get a ranked list: what's worth doing, what can wait, and what to leave alone, with what each would take. The price is fixed and quoted after the first call.

    Example format

    An example of the ranked list the review ends with. The rows are illustrative, not from a client.
    OpportunityValueEffortVerdict
    Draft replies for the support queueHighLowDo now
    Search across contracts and specsMediumMediumCan wait
    Invoice data straight into the ERPHighHighCan wait
    A model in place of the pricing rulesLowHighLeave alone
    A chatbot on the public websiteLowMediumLeave alone

    You keep: a ranked plan, with what each item would take

  2. Prototype

    Prove it on real data. We build the strongest item on that list on real data, with measurable evaluation. If it passes, it becomes the core of the production system, not a throwaway demo.

    You keep: working code on your data, and the evaluation that proves it

  3. Production

    Make it dependable. Integration with your systems, permissions, logging, monitoring, and load and cost limits. The operational standard you would expect from any system your business relies on.

    You keep: a system that runs in production, monitored, within set limits

  4. Handover

    You own the result. Source code, infrastructure, and documentation are yours. We train your team and stay as long as it's useful, but we build so that you never need us.

    You keep: everything: code, infrastructure, documentation

Three of us. More than fifty years in software.

Two engineers who design and build your system, and an advisor who has run delivery programs at scale. No sales layer, no hand-offs.

Péter DajkaLinkedIn ↗Software architect · Full-stack · AI

Zoltán DomonkosLinkedIn ↗Principal engineer · Java · AI

Gyula PálLinkedIn ↗Advisor · Strategy & delivery

50+

years in software

Built for some of the world's biggest companies.

  • Switzerland's largest bank
  • A European parcel-delivery network
  • One of the world's largest online travel groups
  • A national public-administration portal

What people ask first.

Isn't local AI weaker than ChatGPT or Claude?
For open-ended chat, frontier cloud models still lead. For a specific business task, grounded in your own data, modern open models running on your infrastructure match what most companies need. We prove it with a side-by-side benchmark on your real task before you commit.
Can you work with our existing team?
Yes. Often that's how we work best. We can strengthen your team, take responsibility for one system, or deliver the whole project. Nobody gets replaced, and your engineers keep what we build together.
Do you build with AI yourselves?
Every day. It's part of how we specify, write, test, and document software.
How does an engagement start?
With a free 30-minute call with the engineers who'd build it. If it's a fit, a fixed-price AI Readiness Review. The build after that is scoped and agreed up front: never open-ended billing. Not ready to talk yet? The free AI Readiness Check takes about four minutes, needs no email, and shows your result straight away.

Tell us what you need to deliver.

Start with the free AI Readiness Check: about four minutes, a profile in six areas, and a next step. No email needed to see your result.

Take the 4-minute AI check
Top