aricoma logo avatar

#1 in Enterprise IT

Experience with the use of artificial intelligence in our projects and type solutions

Using AI in the areas of damage detection, predictive analytics and process automation and optimization, Aricoma specialists help companies improve quality, increase efficiency and optimize operations across multiple industries.

Published since: 04. 11. 2024

aricoma avatar
Inspiration

Areas of artificial intelligence that we use

Several areas of AI models and algorithms appear in our projects and our type solutions.

Deep Learning [DL]
An artificial intelligence function that mimics the functioning of the human brain in processing data and creating models for use in decision making.

Computer Vision [CV]
Computer vision algorithms attempt to understand an image by decomposing it and studying different parts of objects.

Machine Learning [ML]
Machine learning allows a machine to make decisions based on past experience.

Neural Networks [NN]
A series of algorithms that attempt to recognize relationships in a set of data through a process that mimics the functioning of the human brain.

Natural Language Processing [NLP/LLM]
The Natural Language Processing model is the science of reading, understanding, and interpreting language using a machine.

Cognitive Computing
Algorithms that attempt to mimic the human brain by analyzing text, speech, and images in the way a human would.


Detection of the level of road damage


We have developed a neural network for recognition and analysis of laser images of roads. This [DL/CV] AI model detects microscopic surface damage and provides accurate data for repair and maintenance planning.

The solution significantly reduced data processing time and enabled more efficient resource allocation and long-term cost savings.


Automatic defect detection on conveyors


Within automotive manufacturing, we have implemented an advanced [DL/CV] model that monitors conveyors in real time and detects dozens of types of damage, including cracks and wear.

The system immediately informs maintenance of the need for intervention, which has led to reduced unplanned downtime, increased productivity and improved the overall reliability of the production process.


Optimisation of production processes


For Schwan Cosmetics, we deployed an advanced machine learning [ML] solution that monitors the cosmetic pencil manufacturing process and predicts deviations from the plan. The system detects potential problems, such as downtime or increased scrap production, and offers recommendations on how to address them. 

As a result, production throughput has been accelerated, costs per job have been reduced, and overall production efficiency has increased.


Sorting automation 

of emails


We deployed an [NLP] solution for the customer service center of an energy company that automates the sorting of emails into more than 40 categories, such as billing or technical questions. 

The system processes over 2,300 emails per day, which has sped up request triage, increased triage consistency and reduced errors. This has improved overall customer satisfaction.

Optimisation of working standards


For one of the leading retail chains in the Czech Republic, we optimised working standards using advanced location data analysis and the [ML] model. Based on the data analysis, we adapted the working standards to the real conditions in the stores. 

The result was higher work efficiency, faster movement of goods and reduced operating costs.

Using AI to work with documents / custom content


For a company handling a large number of requests, we helped improve the quality and speed of legal services. Using so-called RAG techniques, we combine generative AI with query categorization and relevant information retrieval. 

In this case, we used cloud-based Microsoft Azure AI for [LLM] algorithms. However, the solution can also be offered in our own closed on-premise model.


A platform for creating intelligent assistants


Our Type #1 solution using advanced [LLM] models provides personalized, real-time support. 

The combination of LLM, finetuning and RAG (Retrieval-Augmented Generation) technology enables accurate responses and fast retrieval of relevant information, this increases process efficiency and offers tailored support.

KNIME, the universal platform for data analysis


Our Type #2 solution, using the open-source KNIME Analytics Platform for data analysis and visualization, facilitates business decision-making. 

The platform makes it easy to cleanse and transform data, create machine learning models, analyze text, and create visualizations and reports. This supports data projects across multiple industries without the need for programming.




Intelligent Image Recognition Platform


Our Type #3 solution includes an anomaly detection module that identifies unspecified defects and irregularities based on deep learning [DL/CV]. The system, trained on error-free images, allows anomaly detection without having to define all possible defects in advance. 

Thanks to intuitive annotation, objects can be easily located and the presence or absence of components can be detected.


Share

Take advantage of our AI competencies!


Interested in more information or a quote for your specific situation?

By submitting the form, I declare that I have familiarized myself with the information on the processing of personal data in ARICOMA.

You May Also Like

KEEP IN TOUCH

Subscribe to our newsletters so you don't miss anything important.

By entering your e-mail, you agree to the terms of personal data protection.

We care about your privacy

We use cookie identifiers that are necessary for the functionality of the website. By clicking the "I understand" button, you agree to the use of other types of cookies for statistical traffic measurement or content customization and marketing targeting according to what interests you.

The consent granted in this way can be revoked at any time later or adjusted according to your preferences. This is offered by the "Detailed settings" option, where there is also more detailed information on the whole issue.

I understand Detailed settings Reject everything