A general AI model isn't enough? We'll train and fine-tune one precisely on your data
- We choose the model and approach that makes technical and economic sense for your brief
- We train it on your data – texts, images and figures from production and sales
- We deploy it into operation and set up how it keeps its quality over time
Prompting, RAG, fine-tuning, or custom training?
Prompting means setting instructions for a ready-made model without touching its weights – the cheapest step, sufficient for general tasks. RAG (retrieval-augmented generation) connects the model to your documents and databases so it answers from verified sources, but the model itself learns nothing. Fine-tuning adjusts an existing model on your data so it keeps your terminology, style and output formats. Custom training builds a model for a task language models cannot do – quality control from images, demand forecasting, form extraction.
The order is not accidental: each further step costs more and needs more data and more compute. That is why we start with the cheapest approach that solves the task and move to fine-tuning or custom training only when measurements on your data show it is needed. Hallucinations are not fixed by more training – RAG and testing take care of that, as we explain in the questions below.
Why tackle your own model right now
A general model doesn't speak your language
You don't want to send company data out
Training and fine-tuning need compute
Sound familiar?
You do quality control by eye
You plan by guesswork, not by data
You retype invoices and forms by hand
You don't know which model and approach to choose
Why train your model with Mediatoring
Approach chosen by the task, not by fashion
Prompting, RAG, fine-tuning or custom training – we decide by what the task and your data actually need, and we tell you when your own model makes no sense. We recommend the cheaper route before you pay for the expensive one.
We measure before we deploy
Every model is compared with the baseline solution on a test set built from your real data. You know the accuracy before go-live, not after – and deployment does not start without a number you have seen.
Ten years with AI, over twenty-five years with business data
We have worked with artificial intelligence since 2016 and with e-shops, ERP and websites for more than a quarter of a century. We know what company data looks like in practice, where to clean it and how to connect the model where it is actually used. We have the computing capacity, including access to the LUMI supercomputer.
Research, not just practice
Founder Michal Kubíček researches at Silesian University how language models work with information about companies, and he is the author of books on SEO and optimisation for AI. He leads one of the largest Czech AI communities.
Who builds it
He has worked with artificial intelligence since 2016 and in online business for over twenty-five years. A doctoral student at Silesian University in Opava (brand representation in the outputs of language models), author of SEO books published by Computer Press and of books on optimisation for AI, columnist for Patriot magazine. He leads model training and fine-tuning personally with the Mediatoring team. He writes at kubicek.ai.
They trust us
Continental · Hyundai Motor · Groupon · Union of Towns and Municipalities of the Czech Republic · Technology Agency of the Czech Republic · Datasys · Spokar · Slovak Athletics Federation · Energy Centre of the Ústí Region · VŠÚO Holovousy
What we can train and fine-tune for you
| Type of service | What it includes | Benefit |
|---|---|---|
| Prompt tuning and RAG | Connecting a language model to company documents and databases without training | Accurate answers from your own data |
| Fine-tuning a language model | Tuning to your terminology, communication style and output formats | A model that writes and answers like your company |
| Computer vision | Image recognition: quality control in production, document extraction, object counting | Less manual work and fewer overlooked errors |
| Predictive models | Predicting demand, stock levels or breakdowns from historical company data | Planning by data instead of guesswork |
| Model selection and deployment | Comparing commercial, open and local models by cost and data requirements | A model that fits both the task and the budget |
| Training on the LUMI supercomputer | Custom training and more demanding fine-tuning where ordinary GPUs are not enough | Compute without investing in your own infrastructure |
4 steps to your own AI model
Data and brief analysis
Model selection and approach
Training, fine-tuning and testing
Deployment and further tuning
How we work together
Intro call
Proposal and quote
First working version
Operation and growth
What it costs
Prompt tuning and RAG
Custom model fine-tuning
Training and operation
When your own AI model pays off and what to expect from it
Frequently asked questions about AI model training and fine-tuning
How much does training or fine-tuning an AI model cost?
The price depends on the scope of the task, the size of the model and the amount of data we work with – fine-tuning a smaller model for a specific task costs something different from training a larger model on domain-specific data. During the initial consultation we go through your brief and prepare an indicative tailored quote, so you know what to expect before we start the project.
Which models and technologies do you work with?
We work with open language models (such as Llama or Mistral) as well as commercial APIs like GPT or Claude. Besides language models, we also train computer vision models and predictive models over business data. For more demanding training and fine-tuning we use the computing power of the LUMI supercomputer, one of the most powerful in Europe.
Who owns the trained model and the data we work with?
The model and all the data it is built on remain yours. We share nothing with third parties and never use it to train other projects – we work purely as a supplier on your infrastructure or in an environment we agree on in advance.
Can you connect the model to our ERP, accounting or e-shop?
Yes, integration with existing systems is a normal part of our projects – whether it is an ERP, accounting software, an e-commerce platform or internal databases. The model can then work directly with your real data and fit into the processes you already have in the company.
How does training or tuning a model work in practice?
First we clarify the goal together and go through the available data. Then comes data preparation and cleaning, the actual training or fine-tuning, testing on real scenarios and tuning based on the results. Finally we deploy the model and hand it over together with documentation on how to work with it.
Who benefits from investing in their own AI model?
A custom model makes sense for companies that work with specific or sensitive data, need to solve the same type of task repeatedly – such as quality control in production, document extraction or demand forecasting – or run into the limits of general tools like ChatGPT. If your task is more of a one-off or general nature, an off-the-shelf solution may be faster and cheaper – and we will tell you so honestly during the consultation.
How does communication work during the project?
We agree on regular consultations where we show interim results, discuss any changes to the brief and answer questions. If you prefer, we can also keep in touch continuously by e-mail or video call, whichever suits you.
Can the model be extended or retrained later?
Yes, there is no need to rebuild the model from scratch if your needs change or new data comes in. We can fine-tune it, extend it with new capabilities or retrain it on more recent data, so it grows with your company.
Who takes care of running, backing up and securing the model?
By agreement we can also run the model after deployment – including backups, monitoring and security measures. If you want to handle operation in-house, we hand over the model with documentation so you can manage it yourselves.
Why invest in your own model when ChatGPT exists?
ChatGPT is an excellent general assistant, but many business tasks are of a different kind: you won’t solve defect detection on a production line, demand forecasting or document extraction with it. And even for language tasks, a custom or fine-tuned model has advantages – it keeps to your terminology and style, runs on infrastructure under your control, and at high query volumes works out cheaper than a paid API.
How do you protect data and privacy during training?
Data protection is essential for us – we work only with data you are authorised to use, and we set up the project so that sensitive data never leaves your infrastructure or the agreed secure environment. We comply with the GDPR, sign a non-disclosure agreement (NDA) with all parties, and never use your data to train other models or projects.
Can you improve or fine-tune a model we already use?
Yes, you don’t have to start from scratch. If you already work with a model and are not happy with the results, we look at exactly where it falls short and propose whether fine-tuning, adjusting prompts and data, or switching to a different model is the more effective route.
What will you need from us at the start of the project?
What helps most is an idea of the problem you want to solve and access to the relevant data – the documents, databases or systems the model is supposed to work with. We take care of the rest, including the technical setup.
How do I arrange a free consultation?
Just write to us via the contact form on this page, call +420 555 333 158, or simply pick a slot for a 15-minute call in the calendar. We get back to you within one working day and on the intro call we discuss your brief, your data and the possible solutions. Neither the consultation nor the proposal costs anything or commits you to anything.
Will training a model stop the AI from making things up?
Additional training alone won’t; hallucinations are inherent to language models. Reliability is handled differently: by connecting the model to your documents and databases (RAG) so it answers from verified sources, and by thorough testing on real examples. Fine-tuning adds the right terminology and form of answers on top. We combine both according to what the task requires.
Do you also do image recognition and predictions, not just chatbots?
Yes. Language models are only part of our work. We also train computer vision models for quality control or reading documents, and predictive models for planning demand, stock and maintenance.