InnovAnts

Inspiration

We created our concept to convince German Mittelstand companies to the idea of utilizing AI in their business. As these businesses are mostly family-owned and value their business culture, we focus on the areas they are already trying to improve by themselves or other solutions.

What it does

Electricity Usage Optimization

Problem

The transition to renewable energy exacerbates Germany's power cost problem, leading to higher electricity prices. However, businesses can lower costs by leveraging price fluctuations in the energy market. Using AI-driven energy management systems, companies can optimize their energy consumption, purchase electricity when prices are low, and reduce usage during peak times. This approach not only cuts costs but also enhances energy efficiency. Additionally, investing in energy storage solutions allows businesses to store cheap electricity for later use. In summary, our goal is to help the factories spend the lowest cost while maintaining the capacity.

There are indeed lots of competition from those developed companies. However, there are few companies doing energy price optimization with energy storage systems (ESS). That’s the reason we can stand out.

We will cooperate with infrastructure companies and battery companies, we provide algorithms and they provide infrastructure. We will quote to the customers and we will share the profit with infrastructure companies.

Predictive Maintenance

One of the key challenges holding back the German Mittelstand from competing effectively on a global scale is its relatively high operating costs. Instead of disrupting their deeply rooted family-centric company culture or altering established management and HR practices, our focus is optimizing production lines and reducing everyday operational expenses.

Depending on the age and condition of their machinery, many Mittelstand companies already rely on preventive and planned maintenance strategies. Modern equipment often includes predictive maintenance capabilities, using sensors and real-time data to trigger alerts. However, these solutions frequently need to be more cohesive and equipment-specific.

Our solution bridges this gap by introducing a centralized AI-powered predictive maintenance platform. This platform is designed to unify and support machinery of various ages and technological capabilities. By integrating data from both legacy systems and modern equipment, our solution delivers actionable insights to optimize maintenance schedules, minimize downtime, and extend the lifespan of machinery. This ultimately drives down operating costs and enhances overall efficiency.

How we built it

Electricity Usage Optimization

Key Features

Combining power price forecasting and power consumption prediction to optimize the decision of energy procurement. The basic principle is to procure energy at a lower price and store it in the batteries. Later on, use the storage power when the price rises. The power consumption prediction enables us to be prepared for enough power before there’s a big demand. In this case, we have to procure electricity beforehand even if the price isn’t low enough. After that, our algorithms will make procurement suggestions for the customers.

Implementation:

Data collecting: Weather data: humidity, temperature, precipitation, wind speed…. History data of electrical pricing Peak and off-peak price History data of capacity History data of the amount of order History data of producing time Electrical consumption for each product Economy Some of the input data are public and some history data have to be asked from the company. The confidential data will be safely stored in our system because we have strong cybersecurity. We can use web crawlers and NLP techniques to fetch public news which can help the company foresee market trends to predict future demand and predict power consumption even further.
Model building: Use a deep learning framework like TensorFlow or PyTorch to build your LSTM model. Define the architecture of the LSTM network, including the number of layers and units per layer.
Model training: Train the LSTM model on the training data. Use a loss function like Mean Squared Error (MSE) and an optimizer like Adam. Monitor the training process and adjust hyperparameters as needed to improve performance.
Combine the prediction with collaborated energy storage systems(ESS).
Give out the suggestions

Predictive Maintenance

We employ a broad range of sensors—including, but not limited to, temperature sensors, motion sensors, and cameras— in addition to the built-in sensors already present in many machines. Unlike traditional IoT solutions, our approach is not limited by fixed device-specific algorithms. Instead, it leverages advanced AI for anomaly detection and triggering actionable suggestions, making it adaptable to diverse operational environments.

Implementation Phases

Model Training Phase: In this phase, we collaborate closely with client professionals to conduct supervised learning for our AI model. This involves gathering and analyzing data specific to the client’s machinery and production processes. While this phase may take longer for initial customers, the model will grow smarter and faster with each new deployment, creating a compounding improvement effect.
Employee Training Phase: Employees previously responsible for quality control and maintenance will be trained to operate the system effectively. They will learn how to interpret alerts, respond to actionable insights, and address any false determinations made by the AI. By the end of this phase, the system will reach full operational readiness.
Operation Phase: Once the system is fully implemented, we transition into ongoing support and after-sales services. This phase ensures that clients receive the guidance they need while fostering a long-term partnership between us and the Mittelstand as a trusted service provider.

Architecture

Cisco’s Industrial IoT portfolio already provides state-of-the-art solutions, leveraging AI in network management. Rather than reinventing the networking backbone, we will focus on integrating our primary product into an existing robust infrastructure.

Depending on the connectivity requirements, the sensor layer will be deployed using Cisco Catalyst PoE switches or Industrial Wireless (IW) access points. These devices will provide seamless integration for sensors such as temperature, motion, and cameras while ensuring reliability in industrial environments.

Data from the sensors will be processed at the edge using Cisco Edge Computing solutions, such as the Cisco IOx framework. This will enable real-time anomaly detection and predictive insights, minimizing latency and dependency on external computing resources. The entire network will be established using appropriate Cisco Catalyst routers optimized for industrial use cases. These routers will provide secure, high-performance connectivity across all network layers.

To safeguard the network, Cisco Secure Firewalls and Identity Services Engine (ISE) will be integrated into the infrastructure. These solutions will protect against external threats while managing secure device and user access policies.

Our company will manage the entire network using tools like Cisco Catalyst Center or IoT Field Network Director (FND). This approach ensures that our product remains an all-in-one solution, eliminating the need for additional IT personnel. It also enables robust cybersecurity through centralized monitoring and proactive threat mitigation since it is managed by a team with cybersecurity awareness.

Challenges we ran into

Cybersecurity in Cloud AI Services

Cloud integration when it comes to AI is essential since running AI models on the edge is usually not practical and cost efficient for companies. That is why over the past several years many cloud providers started to provide services specifically optimised for AI integration.

With many cloud providers being from outside of EU [1], it is hard for European mittelstand companies to trust that their data will be protected. Especially with USA's Cloud Act directly infringing some of the clauses in the GDPR [2], this distrust is warranted. As a startup company that wants to bring AI to the forefront of mittelstand companies priorities, and make this process easier for them, this poses some challenges for us.

Problems:

Companies do not trust cloud service providers with their private data.
Companies have a robustly built system and don't see the need to change.
Companies lack the cybersecurity personel/knowledge to do this transition securely.

After careful research we have decided to have Open Telekom Cloud is the provider that we will use for our services. As a German company and being around for many years, Deutsche Telekom is an established and well respected brand. Also as we can see from Figure 1 it is the dmost used cloud provider in Germany and Europe after the US based companies.

Figure 1:[3]

Open Telekom is a good choice for our case because of how similar the provided services are to the US based counterparts. They have the one to one replacement for most of the services [4]. Even though AWS is GDPR compliant, their project AWS European Sovereign Cloud has yet to launch. AWS European Sovereign Cloud, which aims to bring more sovereignty with data to european users, will launch by the end of 2025 [5]. Until that project is realised and is at a stable state, Open Telekom Cloud remains to be the best option at hand, with regards to market share, customer trust and regulatory requirements.

Another focus we are going to have is the training of the staff that will be interacting with our services. According to 2024 Cloud Security Report by Check Point, 41% of all cyberattacks on cloud have been a result of the lack of security awareness among current employees. This shows how much problems can a simple training given to employees prevent. One of the other services we provide is that for any of the subscribers of our service will receive video and text tutorials, live support in case of any problems and annual trainings and testing to raise the awareness and ability of the employees when it comes to using cloud services.

What's next for InnovAnts

In cooperation with Cisco, Innovants can be the solution to Germany's long-lasting problem of convincing the business-owners to innovate and making them competitive in international scale.