Skip to content

SERVICES

AI and hardware

Software that sees and measures: computer vision, AI that answers from your documents, and devices with sensors and firmware, from the first prototype to a product people use.

Who it is for

Companies that want to check or measure something automatically: a process, a product, a field.

Teams with documents or archives they want to search and question in plain language.

Founders building a device: a sensor, a controller, a connected product.

What you get

  • Computer vision on cameras, drones or phones (OpenCV, PyTorch)
  • AI assistants that answer from your documents and cite their sources
  • Firmware for microcontrollers (ESP32 and others) and the electronics around it
  • The app or dashboard where people see the results
  • A working prototype first, then the product

How we work

  1. A first conversation

    You write a few lines in the project form or book a 30-minute call: what you want to build, for whom and by when. I reply by email with my questions.

  2. Scope and quote

    We agree in writing on what the first version does and does not do, what you receive and what it costs. Nothing starts before that.

  3. Built in stages

    The work is split into short stages. After each one you see the product running and can change course before the next one starts.

  4. Launch and handover

    We put it online, hand over the access and the documentation, and agree on how it is looked after.

How the price is set

Every project is quoted after the first conversation, for its own scope. The price depends on what has to be built, on how many people and systems it touches (users and roles, integrations, devices) and on the deadline.

The quote is split into the same stages as the work, so you know what each one costs before it starts. No figure is given before we know what the project is.

Code, data and hosting

The source code, the data and the accounts (domain, hosting, app stores) are yours, and the code lives in a repository you own.

Personal data is handled under the GDPR: the product stores only what it needs, and when we process data on your behalf we sign a data processing agreement.

After launch

After launch we fix what breaks and keep the software up to date, under a maintenance agreement sized to the product. New features are quoted like a new stage.

Questions

Do you need our data to start?
A sample is enough for a first prototype. The data stays yours and is used only for your project.
Can the AI run on the device, without the cloud?
Often, yes: small models run on the device or on a local computer. The choice depends on the accuracy you need and on the cost of the hardware.

Projects like this

SolCane, Vineyard AI Field and ServiceProof put hardware, AI and the apps around them in one product.

  • SolCane

    A smart white cane for blind people, by Solemtrix. It feels obstacles, sees, and speaks Romanian.

  • Vineyard AI Field

    A vine-by-vine inventory of a vineyard from drone imagery, plus an inspection route on foot. Built at DeepTech GigaHack 2026.

  • ServiceProof

    Two cameras at the car lift, a custom hardware controller and AI video analysis that documents the work done on every car.

  • SeraFullAutonoma

    A greenhouse that monitors itself, makes decisions and talks with the people who run it. My bachelor’s thesis, graded 10/10.

  • Civic AI for Timișoara

    AI summaries of Local Council decisions, and an agent that answers from digitised archives and cites its sources.