Practical AI solutions for real workflows

AI that does a specific job well: reading documents, recognising images and handwriting, understanding text and predicting what comes next.

What we build

  • LLM-powered workflows and assistants built on large language models
  • OCR: turning scanned documents and photos into structured data
  • Image and text recognition
  • Handwriting recognition
  • Predictive analytics on your own data
  • Natural language processing: classification, extraction and summarisation

Our approach

We start from the task, not the model. First we define what the system should do and how we'll measure whether it does it well, then choose the tools: a hosted model through the OpenAI API, or our own pipeline with OpenCV, TensorFlow, PyTorch and Hugging Face. The result plugs into your existing web or mobile product.

Background and related work

Jozef holds a master's degree in artificial intelligence focused on neural networks and image recognition. In practice we've built OCR and invoice recognition into banking apps for Zentity, and app architecture for AI-powered features in Syngenta's crop protection apps. See the case studies

Common questions

What kind of AI projects do you take on?

LLM workflows, OCR, image and text recognition, handwriting recognition, predictive analytics and natural language processing.

Which AI tools do you use?

Python with the OpenAI API, OpenCV, TensorFlow, PyTorch and Hugging Face, depending on the task.

Can AI be added to an app we already have?

Yes. Because we also build web and mobile software, we can connect an AI component to your existing product.

Tell us what you're building.

Send a short description of the idea, where it stands today and when you'd like to launch. Jozef will get back to you personally.