How do I start an AI project in my company?

By analyzing existing data and processes. This is followed by a proof of concept (4–8 weeks) to assess potential and economic viability.

What are the typical costs of AI development and operation?

A proof of concept usually costs CHF 20,000–60,000, while productive systems including data preparation and operation cost CHF 80,000–250,000.

How do I measure the success of an AI solution?

With clear KPIs such as accuracy, throughput, error rate, or ROI. A comparison before and after implementation shows the actual benefits.

How do I integrate AI into existing systems?

Via APIs or microservices, e.g., connection to ERP, CRM, or production control. This preserves the existing IT landscape.

How long does it take from introduction to productive use?

A pilot usually takes 6–10 weeks, a productive system with integration 3–6 months.

What architecture makes sense for AI?

Cloud models are suitable for scaling, edge solutions for real-time processing. Hybrid approaches are often used for flexibility.

When is AI economically viable?

When repetitive, data-intensive processes can be automated. The ROI often becomes apparent after 6–12 months of operational use.

Services

Digitalisation Artificial Intelligence Blog

How does Artificial Intelligence work?

How can I support you?

Robert Schmuck
CTO | Technology

+41.. Show number +41 55 253 00 53

Get expert advice. Discover our success stories.

Artificial intelligence works by enabling software to learn from data, recognize patterns, and make decisions independently.

In this article, you will learn:

AI learns from experience: Instead of rigid rules, AI uses data to recognize connections and make predictions.
Deep learning goes deeper: Neural networks imitate human thinking and solve complex tasks such as speech or image recognition.
Practice instead of theory: With a clear database and targeted modeling, soxes implements AI projects that deliver real efficiency gains.

Artificial intelligence and soxes

What is AI? Artificial intelligence (AI) is changing the way we work, and soxes is at the forefront of this change.

By using AI to automate processes, analyze data, and improve decision-making, soxes helps companies become more efficient and competitive.

AI applications can, for example, help companies make more accurate sales forecasts, calculate optimal inventory levels, and achieve significant efficiency gains.

If you want to leverage the benefits of AI for your business, we are the right partner for you. Develop your personal AI strategy with us now!

What is artificial intelligence?

Artificial intelligence describes systems that perform tasks that previously required human thinking. These include recognizing patterns, understanding language, making decisions, and solving complex problems.

AI is not a single program, but a collective term for various approaches and technologies that analyze data, learn from it, and continuously improve.

The basis for this is machine learning (ML): learning from data.

How does machine learning work?

Machine learning is the process by which software learns independently from examples instead of acting according to rigid rules.
1

Collect data

Historical or real-time data forms the basis.
2

Preprocessing

Remove noise, normalize values, select features.
3

Training

Algorithms recognize patterns in the data and adjust their parameters.
4

Validation

The model is tested with new data to check its accuracy.
5

Deployment

After successful testing, the AI makes predictions or decisions independently.

Deep learning and neural networks

Machine learning can be used as part of a software solution to automatically learn from data sets and improve software performance for specific tasks. To this end, specialized AI algorithms are used to analyze data and make predictions.

The best-known area of artificial intelligence is deep learning. This involves attempting to mimic the functioning of the human brain with neural networks. These networks are trained with large amounts of data: first, the data is collected and processed, then it is used as training material for the network. The end result is a model that is capable of making predictions about new, previously unknown data.

Practical examples of the use of deep learning include virtual assistants, facial recognition, translations, and the vision for driverless vehicles.

Example: A deep learning model for quality control can learn from thousands of product images to detect subtle deviations that escape the human eye.
This can massively reduce the error rate in production.

How soxes uses AI in practice

Data analysis

We check data quality, sources, and potential.
Modeling

We select suitable ML or DL models and train them on real data.
Integration

The AI is integrated into existing systems or workflows.
Evaluation

Ongoing optimization and monitoring ensure long-term stability.

Sam Picek, Lead ML/AI Engineer soxes AG

Opportunities and challenges of AI

The use of AI offers enormous opportunities. At the same time, it requires realistic planning and a clean data foundation.

Opportunities:

Increased efficiency and reduced workload through automation
Faster and more objective decisions
New business models based on data-driven insights

Challenges:

Quality and availability of data
Data protection and IT security
Change management and employee acceptance

To successfully implement AI, you need not only technology, but also an understanding of the processes, people, and goals within your company.

AI prototype for error detection

1 Project Overview

Emmi, one of the world's leading manufacturers of dairy products, uses state-of-the-art technology in its production facilities. soxes developed an AI prototype for Emmi for automated error detection in the yogurt production process.
2 Challenge

Yogurt packaging must be flawless so that products can be correctly labeled and further processed. The goal was to reliably detect deviations in real time.
3 Solution

The AI application processes images of packaging captured by cameras. Using Microsoft Custom Vision and a Python app, the images are analyzed and compared with product data. Deviations, such as an incorrect lid, are immediately detected and reported.
4 Result

Within six weeks, an efficient solution was developed that ensures production quality in real time. Faulty products are identified at an early stage, reducing waste and increasing process reliability.

600+ SUCCESSFUL PROJECTS

400+ HAPPY CUSTOMERS

130+ TEAM MEMBERS

24+ YEARS OF EXPERIENCE

How does soxes use AI to transform software development?

At soxes, we use our expertise in artificial intelligence to develop software solutions that are smarter, more efficient and more user-friendly than ever before. We specialise in a wide range of AI technologies, including natural language processing, machine learning and computer vision. Our team of AI experts works closely with our customers to understand their individual needs and develop solutions that fulfil their specific requirements.

How does soxes use natural language processing to improve communication?

One area where artificial intelligence is having a big impact is natural language processing, which is about teaching machines to understand and interpret human language. With the help of natural language processing, we can develop software solutions that are more user-friendly and accessible, allowing users to interact with the software in their natural language. This can include voice-controlled assistants, chatbots and other tools that facilitate communication with the software and provide the information needed.

How does soxes use machine learning to automate processes?

Another area in which artificial intelligence plays a major role is machine learning. This involves teaching machines to learn from data and make predictions based on this data. Using machine learning, we can develop software solutions that are smarter and more efficient, enabling organisations to automate processes and reduce the need for human intervention. This can include everything from predictive maintenance to fraud detection and risk analysis.

Specific use cases for AI

Here you will find exciting customer projects that show how data, technology, and ideas come together.

Discover where artificial intelligence is already having a practical impact today.

Go to Article

Frequently asked questions

How do I start an AI project in my company?

By analyzing existing data and processes. This is followed by a proof of concept (4–8 weeks) to assess potential and economic viability.
What are the typical costs of AI development and operation?

A proof of concept usually costs CHF 20,000–60,000, while productive systems including data preparation and operation cost CHF 80,000–250,000.
How do I measure the success of an AI solution?

With clear KPIs such as accuracy, throughput, error rate, or ROI. A comparison before and after implementation shows the actual benefits.
How do I integrate AI into existing systems?

Via APIs or microservices, e.g., connection to ERP, CRM, or production control. This preserves the existing IT landscape.
How long does it take from introduction to productive use?

A pilot usually takes 6–10 weeks, a productive system with integration 3–6 months.
What architecture makes sense for AI?

Cloud models are suitable for scaling, edge solutions for real-time processing. Hybrid approaches are often used for flexibility.
When is AI economically viable?

When repetitive, data-intensive processes can be automated. The ROI often becomes apparent after 6–12 months of operational use.

Ready for your first AI solution?

Work with us to find out where artificial intelligence can really pay off in your company.

Analysis of your database and processes
Initial roadmap with concrete steps for action

Direct number

+41 55 253 00 53

Request AI consulting with soxes now

Vier Personen sitzen am Tisch und diskutieren über Künstliche Intelligenz (AI steht auf einem Whiteboard hinter ihnen)

How can we support you with your software project?

Robert Schmuck
CTO | Technology

+41.. Show number +41 55 253 00 53

Please contact me Our References

Expert knowledge for your project

CTO at soxes and driving force behind technological excellence: Robert Schmuck is responsible for architectural decisions, software quality, and AI expertise. He leads the development teams, complemented by a master’s degree in industrial engineering from the Karlsruhe Institute of Technology and Scrum certifications, with a clear eye for processes, efficiency, and strategy.