Faster customer replies
Support assistants, internal knowledge bases, or customer portals that answer from your own documents.
Less waiting, more consistent answers.
MPTR — since 2015
We build it. Then we run it.
Web, native mobile, backend, operations, and AI under one roof. Since 2015, the rule has been simple: whoever builds it is also responsible for running it.
Some of our clients
Innovation for small and medium businesses
You don't need an AI demo. You need less manual work.
If you can tell your company could move faster, but you don't want a technology detour, we start here: the workflow where software or AI frees up real hours soonest.
You won't leave the first conversation with generalities.
You'll leave with 1–2 concrete next steps: what's worth automating, what isn't, which data you'll need, and roughly what cost range to expect.
Less manual admin
Connect quotes, invoice data, documents, emails, and reports with human checkpoints where they matter.
Your team stops spending the day copy-pasting.
Smarter internal search
Turn old files, contracts, manuals, and CRM data into knowledge people can actually find.
Knowledge no longer lives only with whoever remembers it.
More stable operations
Monitoring, backups, incident response, and security basics before growth exposes weak points.
Fewer surprises in production.
Why us
We don't just hand it over. We stay accountable.
A good system proves itself after launch: when traffic grows, the business changes, or monitoring raises an alert.
That's why the same team designs, builds, and operates it. We host, monitor, patch, and fix. When something goes wrong, you don't end up in a detached support queue; an engineer who knows the system from the inside looks at it.
AI belongs in the same operating model. It's not a side demo. It's another part of the product, with data, permissions, cost, failure modes, and security. If we do it right, the system removes work instead of creating more.
What we do
Three pillars, one team.
01
Design & Build
UI, mobile, web, backend. Our own code. Plenty of teams can write software. Fewer write software that's pleasant to use and stays stable in production. That's what we build.
- Native iOS and Android apps
- Astro, Next.js, Vue — server-rendered, fast web
- Postgres, Node, Go — APIs and backends that honor your existing systems
02
Run & Secure
If we built it, we run it. No separate support team taking over. When something breaks, an engineer who knows the code looks at it, so the cause is found faster. Obvious, yet rare.
- Continuous monitoring, business-hours response, custom SLA if needed
- GDPR, NIS2, pentest, incident response
- Our infra or yours — Cloudflare, AWS, self-hosted
03
AI-native
AI that does real work. Claude and GPT in your product, or Qwen3.8 / DeepSeek V4 on our own hardware. Agentic workflows with MCP, OpenClaw-based local-first agents, AI security. In live systems, not demos.
- LLM integration: RAG, assistants, document data extraction
- On-prem inference in Budapest — data stays in the EU
- AI agents and MCP in our own daily work too
One project
A system we still look after today.
Logistics / Transport
Tacho561 — tachograph assistant and fleet manager
Driver mobile app, dispatcher web portal and cloud backend for hauliers: driving-time tracking, Bluetooth tachograph download, digital driver log and CMR, in line with AETR and Regulation (EC) 561/2006.
App + portal
and backend from one team
561/2006/EC
and AETR compliance
9 languages
on the website
AI in production
AI inside your product,
not in a side demo.
Experimenting isn't the goal. RAG, assistants, automated workflows, Claude/GPT integrations, or local models go into the product only when they measurably save time, errors, or manual work.
- LLMs in your product: RAG, copilots, search. In production, not as a demo.
- AI assistants and chatbots for customer support, internal knowledge bases, and admin work.
- Agentic workflows and MCP servers, for when AI needs to touch your systems.
- AI security: prompt injection testing, data access rules, model access control.
- Models on our own infrastructure: your data stays in the EU.
Production pattern
/ai · data sovereignty
Run AI close to the data.
Cloud models when they fit. Local or EU-hosted models when privacy, cost, or control matters.
First estimate
Tell us what you want to build. We'll reply in writing.
Based on a few questions we send a short summary: suggested approach and stack, estimated cost range, main risks.
How we work
After launch, we're still here.
Launch isn't the end of the project. We measure, fix, and improve for as long as you need us.
- 01
Discovery
In the first week or two we map the problem, the goals and your existing system. You get a clear architecture sketch, in writing.
- 02
Build
Native mobile, modern web, your own backend. Our own code, not a template site, built around the systems you already have.
- 03
Run
We don't disappear after launch. We host, monitor, patch, and fix it. We respond during business hours, and beyond that under a custom SLA if you need it.
- 04
Improve
Software is never done. We improve it on a monthly rhythm and document every change.
Start
Got an idea, or worried about your system? Get in touch.
The first conversation is 30 minutes and free. By the end you'll know whether working together makes sense, or whether another route suits you better.