Photo - AiSight GmbH
50964

AiSight GmbH

Machine Diagnostics for Manufacturing

Germany
Market: Auto Maintenance, Mechanical engineering, Artificial Intelligence
Stage of the project: Operating business

Date of last change: 18.01.2021
Rating
equalizer from 2000
help
Calculated and estimated occupancy of the project (more about ratings)
My rating
1
2
3
4
5
6
7
8
9
10
Average rating:

Idea

AiSight is contributing to the digital transformation of the manufacturing industry, with the creation of a plugin vibration-based sensor kit that leverages AI to monitor the condition of a machine in real-time, optimize its parameters and predict malfunctions.

Current Status

Four months after founding AiSight in July 2018 we installed the first 20 prototype sensor kits at our first two medium size pilot clients. Currently AiSight has 4 pilot customers running live. The pilot clients include 2 industrial medium size clients, as well as BASF and Miele. The AiSight solution works for machine types including plastic extruders and injection molding machines, as well as rotating equipment including electric motors, pumps, compressors & industrial fans. The customer base includes 12 high profile customers, like Henkel, Audi, Fraport, Gerolsteiner and many more and the customer pipeline is filled with more than 165 interested clients, including many large clients like Honda, Michelin, Toyota, Trumpf, Dräxelmaier, Diehl etc.

Market

Our target customers are in the German Mittelstand and large industrial companies, focusing on the following applications: injection molding, plastic extruders, electric motors, pumps & conveyor belts.

The total market for maintenance services is €140 billion worldwide. Approximately 16% (€22 billion) of this is generated in Germany (Statista, 2018). The average margins in machine services are around 10%. Predictive maintenance promises to reduce the costs for service providers by up to 50%. Even if one assumes a cost reduction of only 10%, a savings potential of almost €2 billion is possible in Germany alone. With our business model, which sees 10% of savings as a potential market, this would be a potential market in Germany of about €200 million. This assumed value corresponds very well with a study by IoT Analytics, which calculates the predictive maintenance market worldwide in 2016 at just under 1.5 billion USD (Analytics, 2017). In addition, an annual growth rate of 39% is expected.

According to a study by PWC, less than 10% of the respondents have implemented Predictive Maintenance in some form, but well over a third of the respondents plan to implement it within the next 5 years. This suggests that the benefits of predictive maintenance are well known and recognised, but that the market is still in its infancy (PwC, 2017).

Initial focus is on the DACH region and Europe. Over time, however, this can be extended to the global market. A study by the VDMA states that approx. 12 % of sales in the predictive maintenance market are generated by start-ups. In principle, the targeted solution can be implemented in any mechanical system and is therefore not limited to a specific sector. The focus market at the beginning is the industrial sector and can later be extended to other markets.

Problem or Opportunity

Machine failures account for 20% of the total cost of production. In Germany alone, this corresponds to € 44 billion a year. These arise on the one hand by wear and wrong behavior of the machine, on the other hand by the change of the machine parameters during operation, which leads to faulty output. Machine failures not only decrease productivity but also impact negatively on maintenance costs. The complete elimination of maintenance is not possible, but the optimization of such processes is definitely a priority in the industry. Thus, the first problem to solve is that of understanding where machine failures originate, gaining a thorough knowledge of their behavior to predict potential malfunctions and schedule optimal maintenance sessions. Another problematic aspect is to bring innovation to processes that are generally very static. Machines in production halls are generally diverse and rarely equipped with sensors and connected to a network. Machines’ operators often still inspect the equipment physically and they are reluctant to send this data to cloud-based services for analysis for data security reasons, or simply because production networks are not designed to handle transmitting large volumes of data. To encourage innovation in such processes, there need to be solutions that prioritize user-friendly approaches.

Solution (product or service)

AiSight develops a solution utilizing artificial intelligence algorithms to determine the state of a machine in real-time, predict errors and dynamically regulate the machine parameters based on sensor data. The solution consists of both hardware and software components.
The hardware is an easy-to-install (AiSight Sensorkit) device that includes a variety of sensors and is optimized to allow Plug & Play installation within minutes without specialized expertise.
The AiSight software uses machine learning models to value the condition of a machine, identify anomalies and determine causes of errors based on patterns and physical parameters in the sensor data. The key algorithm of this solution is "acoustic fingerprinting", in which physical parameters are extracted from the sensor data in order to identify patterns.
The software is based on a self-learning approach: our algorithm that can quickly re-learn unknown anomalies and therefore requires far less training data.

Competitors

Our closest competitors are Augury and Petasense. Augury is a US-based company. Augury focuses only on Predictive Maintenance, without providing operators with further insights to optimize machines’ use and control the quality of their output. Augury relies on a remote processing unit: the raw data gathered fr om the sensors (through two measurements per day) is sent to the cloud and processed there. Before the actual monitoring begins, however, Augury has to gather historical data from the specific machine under control in order to train its models accordingly.
Petasense is a US-based company that offers solutions for predictive maintenance as well as optimization and asset reliability. Contrary to AiSight, each sensor node has a different purpose, there is one exclusively meant to analyze vibrations for maintenance purposes, another that aims at optimizing the machines and finally, the cloud, wh ere all data is stored and analyzed is a separate solution, too. Customers can choose the degree of analysis and the coverage, by selecting different monthly subscription plans. However, a deeper degree of analysis is only available with the highest tier subscription plan, undermining the solution’s scalability.
From a positioning point of view, Augury offers a more integrated solution, but is not so advanced in terms of easiness of installation. Petasense is more advanced under that perspective, but is less integrated.

Advantages or differentiators

The factors that make AiSight different is that our solution provides customers with a very deep machine analysis - enabled by our innovative sensor kit - while being still extremely easy-to-use and cost-effective.

What enables the depth of analysis is the AiSight sensor kit, equipped with high-frequency band up to 15 kHz and high-resolution vibration, temperature and magnetic field sensors. This is not typical for this kind of solution and allows the detection of smallest deviations from the machine’s normal behaviour and of course root-cause analysis.
AiSight’s algorithms rely on pre-trained models for specific machine types, which implies that it is not necessary to re-train the models and wait for 6-month until there is a value. This means you can install the sensor, login to the dashboard and see if a machine already has a failure and know its root-cause.
The data gathered is processed on the device: there is no need to set up a server and the data traffic is reduced to a minimum.
AiSight goes beyond simple Predictive Maintenance, as it focuses on complete machine process optimization by dynamic machine parameter setting & can automate quality control processes by detecting false production from vibration for plastic extruders and injection moulding machines.
Finally, the solution is that simple, that it is possible to bring online a whole factory with 100 machines in one day and the cost is around 1/5 to 1/10 of the Augury solution from Israel, that is probably being referred to.

Money will be spent on

- Increasing Team from 17 to 45 employees
- Install 20000 Sensors to get from 1 Million € to 12 Million € annual recurring revenue (ARR)

Team or Management

Incubation/Acceleration programs accomplishment

HAX Accelerator

Won the competition and other awards

Forbes 30 under 30

Photos

Photo 1 - Machine Diagnostics for Manufacturing

Product Video

5,00
1
2
3
4
5
1 voice
Sign in/Sign up
arrow_back
EN
more_horiz
close
visibility205
star0
Add to favorites
Delete from favorites
share
close
thumb_up2
Like
Unlike
Idea
Current Status
Market
Problem or Opportunity
Solution (product or service)
Competitors
Advantages or differentiators
Finance
Invested in previous rounds, $
Business model
Money will be spent on
Offer for investor
Team or Management
Mentors & Advisors
Lead investor
Risks
Incubation/Acceleration programs accomplishment
Won the competition and other awards
Invention/Patent
Photos
Product Video