Use cases

The start-up MiraNext is a cooperation between the companies CAS, Doringer and neke-neke and aims to significantly simplify and improve everyday radiology practice through the use of artificial intelligence.
Our technologies enable optimized appointment scheduling, automated structuring of findings and efficient referral evaluation.
These innovations help to reduce staff workload and promote a balanced and efficient practice environment.
A central aspect of reducing the workload is minimizing the amount of telephone work.
MiraNext aims to largely transfer standard inquiries to digital online services.
For inquiries that cannot be digitized, we offer the support of an advanced telephone bot to further increase efficiency and reduce administrative effort.

Other products in this portfolio include a teleradiology platform that enables radiology resources to be distributed and used efficiently, as well as a self-service patient portal. AI-based software in the medical environment requires the highest safety standards, accuracy and an understanding of the needs of doctors and their staff. We are pleased to be responsible for the implementation of this exciting project.

MiraNext – AI for Health

Product development / StartUp

AI Development, Dashboard/Web Development, UX/UI, Backend Engineering, API Design, Product Ownership, Legacy Integration and Phone-Bot Development

(BC) customer service in companies is under high pressure. Customers want to receive information, products and services quickly and to the point. Without waiting times. This always requires a high level of resources and personnel, and yet the demands on employees are enormous. This leads to overwork, frustration and, in the worst case, high staff turnover and dissatisfied customers.
To prevent these problems from arising in the first place, we worked with our customer from the energy sector to develop an AI system that analyzes, evaluates and classifies (email) inquiries from customers and automatically forwards them to the responsible department.

Modern language models, so-called LLMs (Large Language Models), which are familiar to many from the ChatGPT application, are used to solve the problem. These deep learning models have an excellent understanding of language for analyzing e.g. emails, attachments, etc., but can also return incorrect results (hallucinations), which is why appropriate security mechanisms must be built in.A good example of AI in daily use, to relieve employees and satisfy customers.

Customer service email bot

Energy supplier / Custom Solution

AI development, statistical evaluations, fine-tuning, Azure deployment

Banks have a lot of internal documents and up-to-date data that are essential for day-to-day business.
Every minute invested in searching, reading and evaluating this data has an impact on efficiency and ultimately also on employee satisfaction.

In this customer project from the banking sector, reducing the workload of employees and the associated increase in efficiency is the top priority. We all know this from our private lives: we are flooded with information and data and struggle to separate the right from the wrong, the useful from the superfluous.

In most companieseople struggle with the same problems. A wide variety of data sources have to be consulted and it is often the most experienced employees who ultimately know the solution.

The aim was therefore to bundle the accessible but scattered information in the company and make it accessible via a central access point. As a solution, we established a specialized AI bot that can answer employees’ queries specifically and up to date. It always references the original documents or their localization if the summary is not sufficient or if additional information is required.

Domain experts can independently configure which content the bot uses as information sources in order to ensure maximum flexibility.

Internal employee Bot

Banks / Custom Solution

AI development, statistical data analysis, model fine-tuning, vector databases, dashboard development

Sure, many people are familiar with ChatGPT prompts such as Write an X-Post like Trump. After all, the underlying LLMs (Large Language Models) were also fed with millions of X-Posts, certainly including thousands from Trump. So you first have to teach the model to write like someone else, and for that it needs data.

Our client would like to translate its product texts, websites, etc. into other languages without losing the company’s tone of voice. The texts should therefore also sound in the corporate language when translated. Challenge accepted!

As is so often the case in machine learning, the data basis is decisive for success or failure. The mantra of information technology input = output applies even more to deep learning and AI.

The first step is therefore to clean up and structure the data and break it down into suitable chunks for the model. Then the model is taught a lot of corporate language and technical terms. Finally, everything is integrated into an existing workflow and made accessible via a plugin. Continuous learning is always important for models, i.e. the possibility for users to provide feedback, which the model then uses for further training and optimization.

Tonality Translations

Lifestyle – Marketing / Custom Solution

AI development, data preparation, dashboard/web development, backend engineering, API design, product ownership and plugin integration

If you ask experts in facial analysis, they will answer with a resounding YES. In fact, the“search” for certain features or markers in the face is an integral part of alternative medicine and TCM.

Everyone knows from their own experience that dark circles under the eyes, for example, indicate a lack of sleep, but of course expert knowledge goes far beyond this and requires years of experience. Together with a lifestyle company, we have made use of this experience to make this knowledge accessible to a wide audience. But that’s not all, this lifestyle data was just one building block that was expanded to include other health and sensor data.

But how do you preserve decades of experience and make this knowledge accessible to a wide audience? With AI, of course. As is so often the case, the first step is to collect data. To do this, we created an online platform on which experts could label images of faces. We were then able to use these labels to train the models. We developed a separate CNN (Convolution Neural Network Model) for each of the selected markers in the face. In cases with too little data, we used the transfer learning technique. The result was an app in the app store with which you could take a photo of your face and receive a corresponding prediction. Other features of the app were the integration of a genetic test, a health bot and the integration of sensor data such as Oura. Garmin or Whoop. A cool thing!

Bionary – Future Health

Lifestyle / Startup

AI development, dashboard/web development, backend engineering, API design, product ownership and bot development

Of course by reducing the workload. The excellent standard of tourism can only be maintained if employees are able to respond to guests’ wishes. Employees need time to do this, and AI gives them that time.

Our tool supports reception staff in their daily activities, such as answering inquiries.
While others are still sending out offers, the users of our tool can already take care of the guests personally.

Together with one of the most renowned hoteliers in Austria, we have developed this tool and brought it to market maturity. Further features are in development and will be added in the course of the year.

AI support in tourism

Tourism / Product

AI development, dashboard/web development, backend engineering, API integration and product ownership

Our innovative AI solution supports you in onboarding new employees! With our intelligent assistant, new team members can immediately clarify all their questions about the company directly and easily. From details about the company culture to specific work processes, our tool ensures that every employee is fully informed from day one.

Increase the efficiency of the onboarding process and promote a committed working atmosphere. Let our AI integrate your team faster and more effectively!

The LLM (Large Language Model) based bot can be individually configured and adapted to the needs of each company. Whether in the cloud or in an enterprise context, we will find the right deployment model for your scenario.

Onboarding bot

Business / Product

AI Development, Dashboard/Web Development, Backend Engineering, API Design, Product Ownership, Legacy Integration and Phone-Bot Development

Stylish, innovative and hygienic product selection, payment and dispensing via smartphone only. Connection via QR code or NFC tag. No additional controls or modules required. Any cladding .

This is what we call meta-operated automata.

If you are interested, just drop by for a coffee in our workshop and be sure to watch our concept video below.

Detailed statistics on products purchased, interactions and hot sales. QR code and photo-based filling of vending machines. Alerts on expiry date, missing products and batch purity. ERP integration, EFSTA fiscalization and digital receipt.

Meta Operated Automats

Vending machine / product

Automated hardware integration, chip design, dashboard/web development, backend engineering, API integration and product ownership