Youmoni Private AI Models and Infrastructure for Thinking Things

Youmoni IoT & Private AI for understanding, predicting and acting.

You will find new insights and predictions with Youmoni Private AI. With data from your products, and customers interactions, AI and deep learning models will be trained and patterns be learned. You can also introduce smart AI based assistance in Apps guiding end customers for solving a failure or a predicted maintenance task.

With Youmoni Private AI you can securely train and learn using data from your own products, content and feedback from your customers and experts, accelerating your customer experience and position in a competitive landscape.

You will learn how your products really work, when they exceed expectations, behave in different situations and why they fail so you can improve and start being proactive.

In short, connected products combined with private AI will give you new insights that will improve your offerings and understand how your customers really use them.

More data will bring more clarity, and it will be your new reality

Business value using Youmoni IoT and Private AI

Connected products combined with Private AI bring new business values.

Minimise down time in industrial production and machines by predicting problems.
Predictions of product behaviour give insights and minimise expensive unexpected maintenance.
Improve product quality and customer experience by reducing hidden errors identified by deep learning from multiple data sets.
Smart digital assistance using privately trained models will improve service performance, minimize training and reduce customer service time.
Understanding your customers usage of your products will improve product development.

What Youmoni Private AI means for you

Private Models trained with your data

Using Youmoni private LLMs and Deep Learning models (CNN, LSTM) training can be more specific and smarter.
The models will be optimized using your sensor-, product data and using your content, product information, service manuals and experts’ knowledge not available for public models to train on.

Private IoT and AI infrastructure in EU

The models will execute and run in the Youmoni AI infrastructure in Sweden / EU. It means that your data is processed locally in Youmoni Servers equipped with private GPUs.
Your product and customer data are not leaked in any public model.

Privacy and security first

Security and protecting your data are Youmonis priority. Youmoni use SUSE AI (https://suse.com/ai) to with ensuring screened and sanitized LLMs and Tensor Flow deep learning models.
When the models are executed, they are monitored and isolated when processed in our GPUs.

Correct and smarter response

Training and deep learning using data from sensors on your products and your own content and will result in more precise deep learning with intelligent insights.
It will also provide better digital assistance and proactive notifications for your customers and service staff.

Youmoni Private AI for Assistance

Connected products often come with an end user app. It can be an app for a car, a boat, an industrial machine construction equipment, heater, AC or similar. By training Youmon Private AI with sensor data, content and community insights, personal digital assistance for a specific product can be made available in Youmoni Product Apps or custom apps using Youmonis components and APIs. The AI assistant can precisely proactively guide a user, to get the most out of the product, crating a unique customer experience.

Youmoni & SUSE for private IoT and AI use cases

Youmoni and SUSE works closely. Together we deliver a combined SUSE infrastructure and Youmoni IoT & AI solution for private IoT and AI use cases. Youmoni has established one of the first European IoT & AI Platform hosted in Stockholm, Sweden. We can quickly start and deploy private IoT & AI Solution clouds for customers that want to be secure and based in the EU and not dependent on public clouds or non-EU based infrastructure. For some business verticals it's necessary to be local and private, e.g. in medic-, public-, energy-, security-, and military. The combined Youmoni and SUSE IoT and AI platform with SUSE infrastructure can also be deployed on premise in our customers IT infrastructure.

Click here to read more about Youmoni & SUSE Solutions for private IoT & AI.

How it works — from data to intelligence

How the Youmoni IoT Platform and AI models works. From sensors to data to intelligence.

01. IoT Data Capture

Capture data by using Youmoni IoT Sensor cards, like the Youmoni Sense, and or edge computers that with onboard sensors or integrated data from interfaces like Modbus, RS232, CAN, etc. A lot of data are captured and processed in real time.

02. Preparation & Cleaning

Using the Youmoni IoT Platform services for sensor data processing, the data is classified, transformed and time stamped and stored with your meta data. Prepared for deep learning.

03. Storage and Visualization

Captured and classified data can at any time be visualized in the Youmoni Console or in Customized Apps or other user interfaces. Real time in memory data is available from the platform’s services APIs. Historical data can be retrieved using the API too.

04. Timeline

Thresholds, trends, and other discrete sensor data discoveries and or customer interactions are processed in real time and written to the connected products Timeline. It becomes logbook over the products history and how it behaves.
The Timeline can be read by the Youmoni AI as one source of training and are also used to write discoveries from the AI and or Deep Learning models e.g. a discovery of a detected anomaly.

05. Model Training

The Private Deep Learning, e.g. unsupervised CNN or LSTM models using TensorFlow deployed in Youmonis Private AI infrastructure uses the tagged sensor data from the time series database to learn the products behavior.
Once learned the Autoencoder model LSTM can in runtime detect e.g. anomalies. Customized and supervised training with feedback loops can also be implemented and trained.

06. Youmoni Private Deep Learning Models

Youmoni has feature engineered generic and specific models for predictive maintenance that can learn from multiple sensor data streams e.g. a combination of vibrations, sounds, energy usage, temperature and humidity.

07. Notifications

Any discovery detected that is written to Timeline that is relevant for a target audience can be sent as notifications, dashboard updates and or integration events to external system such as your ERP, CRM, SCADA, Field Service or other relevant system.

08. Deployment and Infrastructure

Youmoni used LLM and Tensor Flow models are screened using SUSE AI and deep learning performed in Youmonis Private infrastructure and NVIDIA GPU clusters.

09. Monitoring and Evaluation

All models are and output are monitored using SUSE AI making it possible to detect problems and or errors made by the models making it possible to evaluate and relearn and optimize over time.

Youmoni Data Flow from Product to Notification and AI in Parallel

Real World Example

A pump motor is placed in the basement of a building and shall pump water if the basement is flooded. The pump motor is connected to Youmoni IoT & AI Platform via Youmoni Sense IoT card. The Youmoni Sense card feels the engines vibrations, listens to the sound of the engine and has a connected temp sensor, a flood sensor on the floor and a relay that controls the electricity to the motor.

The sound and the typical vibrations and typical pump events have been learned by Youmoni Private AI during model training. Every time the floor is flooded Youmoni IoT data capture monitors the flooding status, measures the vibrations and sounds from the motor and the motors temperature. The flooding pattern repeats quite regular, every month and the pattern verified by the learned AI/ML model.

One morning the motor starts, but the flood sensor reports no water and the motor speeds with a higher rpm (casing it to be warm). It has no water to pump. The AI identifies the anomaly and Reacts; It shuts down the motor and reports the action to the Timeline service and to be viewed in the property's online dashboard. An app push and SMS are sent to the Youmoni Service App used by the service organization that the basement pump needs maintenance.

Business Results

No flooded basement
Rescued engine
Planned maintenance

The result is lower maintenance cost. The pump was rescued from overheating and the service organisation can react, plan a service. With the data from the IoT platform the service organisation knows that it’s the flood sensor that needs maintenance and can order it. If the event was not detected at all, the basements would next time have flooded, causing major damage for 10000+ Euros.

Where Private AI makes a difference - Youmoni's focus areas

Industrial operations

With data streams, integrated to the Youmoni IoT Platform, from edge computers, PLC and or Youmoni IoT GW and Sensor cards our Private AI can be trained and learn how normal operations are. If changes are discovered it will be identified and notified
With supervised learning it can also tell what has happened and or what is likely going to happen. It can apply to any machine, engine, pump, fan, ventilation system or complex combined systems.
The result is less down time, less unexpected maintenance and minimized and expensive failures to repair and manage. It will improve quality, save time and money.

Energy Management

Data streams from Youmonis integrated energy and effects sensors will store and classify a lot of data from industrial machines, boilers, AC coolers and or heaters in buildings.
The Youmoni Private AI can learn usage patterns and energy price predictions combined with logged energy and effects patterns with the result that it can recommend and or control when equipment that consumes a lot of energy should run.
The result is lower energy bills and lower effect peaks minimizing costs and load on grid.

Properties - People Presence & Patterns

With Youmoni Presence sensors with radars and IR the Private AI can learn typical patterns for moving items and people in industrial processes and buildings.
Typical use cases are data input to energy optimisation models, security and or safety applications in properties and public spaces.

Environmental Monitoring

Air and water pollution monitoring with data from multiple sources. Optimizing water usage in golf and agricultural applications. These applications need data from external sources combined with sensor data.
The data streams can be used by the Youmoni Private AI learning complex systems like weather forecasts combined with sensor data from air and soil in agriculture, golf and or properties applications.
The models can be used to e.g. predict when its optimal to add extra water when golfing, giving better results and reducing water usage, framing applications or running ventilation systems in budlings to lower energy consumption.

Youmoni IoT Sensors and edge computing with AI & ML

Youmoni deep learning models can be deployed on the edge in edge computers, e.g. running Youmoni Edge platform with SUSE SLE Micro Linux and Youmoni Platform in containers or on Youmoni IoT cards using the Youmoni Embedded Platform based on FreeRTOS. Once some models have been trained in our central infrastructure and deep learning clusters with GPUs they can use “TensorFlow Lite - TinyML” and deployed at the edge for fast detection of specific data and sensor patterns, e.g. radar, vibration and sound patterns.

A closeup of Youmonis Space Invento HV3 chip

Why Youmoni Private AI

Feature Engineering

Work with our data scientists. Analyze the possibilities, connect and integrate a product and train models to gain new intelligence.

Data from your products

Integrated with Youmoni IoT Platform, edge computing and IoT sensor cards.

Privacy by design

Models deployed in EU/Stockholm in Youmonis infrastructure using SUSE AI.

Faster Results

Ready to use generic models and some specific models, IoT sensors and tools.

Easy integration

With Youmon IoT Platform it’s easy to connect to your existing products and machines with its, data sources and sensors.