Scalable industrial software with AppStore like Purchase process — sounds impossible? Not anymore.



Many people talk about the IIoT, but in fact, there are very few cases, where a real IIoT system has been fully implemented. These projects are rarely pass the phase of Pilot application. Industrial software takes ages to develop and more ages to integrates. It is always long, unpredictable and expensive. Most of the companies do it on their own building development centers or Outsourcing this job. Why does it happen? The answer is that the solutions, one develops for a certain Plant are rarely scalable. And when one cannot scale a solution, it either becomes too expensive for one Plant or the solution one develops is not marginal for the developer. The reasons is that most of the Plants do have different IT architectures, different Automation layer equipment and these leads to the problem above. A separate topic is the purchase process, which lasts years and increases the CAC for the developer even more. Is there a way to make the industrial software scalable and easy to consume for the Productions? The answer was NO for many years. But these days, there is one solution at the market, which changes the status quo and now the answer is POTENTIALLY YES, it can be easily consumed and scaled. But the devil is like always in details. Let’s try to get through.

N1M1 business model in the Industrial context.

A famous “burger” digital business model has been widely described by many books and papers. The idea is simple, in order for something to be scalable, one needs all four layers of the Burger model working together and there can be completely different Players at different layers. I talk about Clients (N), Marketplace (1), Products (M), Infrastructure (1). A good example is Apple. Clients having access to AppStore, being able to install numerous products from different suppliers having a more or less unified infrastructure — iPhones etc. Another good example is Amazon, with it’s store, AWS services and millions of products across the Globe. Obviously, all layers need to be in a way standardized, to allow for seamless communication and clarity in developing the apps, being able to communicate though all the layers of the model. In case of the IIoT this is obviously not the case.

N1M1 digital business model concept

Industrial challenge for the N1M1 is that all infrastructures are built in a different manner, using different approaches and different technologies. If one would want to develop products, he would immediately face tons of questions with hardware and software connectivity, security, scalability, data flows and many other issues. This increases CAC and does not improve the LTV. A bad combination for startups and companies, that want to develop IIoT use cases. Access to the Plant’s problems is also quite limited, so that only Plant managers are aware of real problems they face in the production floor. Under this restrictions and complexity only few companies continue developing the use cases for Production, while the solutions cost too high in most of the cases. Is there a way out of this loop? In fact — yes.

But before I start talking about the solutions, let’s introduce some limitations to what I write about.

Different industries value chains. I like the classification of different industries by the value chains. These are Process manufacturing (e.g. Chemicals, metals, Pulp & Paper, Consumer packaged goods, Fashion and Life science) and Discrete Manufacturing (e.g. Automotive, Aero Space, Industrial Machinery, Electronics and Semiconductors). Process manufacturing is divided into Asset Oriented Value Chains (AOVC) and Brand Oriented value Chains (BOVC). IIoT technologies need to be able to support all types of value chains to be more efficient, but the practice shows, that while Process manufacturers are mainly concerned with the Stable quality of the output product and the Stability of the production line and need tools for Predictive asset maintenance and Quality control. A Discrete manufacturers are more concerned with the Quality of the produced items and Suppliers sustainability (in a broad sense). All these solutions are widely available in the market, one can easily buy stand alone Asset monitoring solutions (e.g. vibration diagnostics or motor current signature analysis etc.).

Problems observed at the Production.

Here comes the first problem, all solutions, despite the production type, are asset specific and need to be integrated with existing IT/OT systems, sometimes with systems at other levels, e.g. PLM, MES, CRM, ERP and so on. These solutions would also deliver a piece of their own IT landscape into already existing and many cases already very complex IT landscapes. Integration takes a lot of time and resources, which at the end creates high switching costs for the Productions. As a result this leads into very complex Purchase processes by productions, increasing the CAC of the solution builder, but also creating quite high alternative costs for the Production itself. One cannot just run 10 solutions in parallel, select the one which fits purpose the best and delete all others. In every case — pilot, purchase, adaptation, integration etc. This is a “samsara” wheel for the Production managers / Plant managers / IT, since they need to manage a huge park of different Apps, that all use different technologies, have different business models and license types, have different quality of documentation and support and “deepness” of integration.

Second problem is corporate data. Job of the IT team is to make sure, that all data from PLCs / other devices is stored securely and reliably, meaning data is not lost, data is backed up and access to data is structured and controlled (of course it is a bit more complex than this, but lets stay at this level of complexity). IT does not know the substance of the data. So when the solution integrator comes and asks for 100Hz spectrum data from the motor case vibration, IT cannot point him to the pace in the data lake, since quite often all data is in the unstructured data lake with a small structured instances in the excels and separate DBs of the Apps / employees themselves. IT has no understanding of data, it’s not their job, while guys working with the automation layer, they know which data is produced by which asset, they can open up the PLC and they can create a required data stream, with a required frequency and they know where to find it in the unstructured data lake. So at the end, the App will create it’s own small substance of structured data, required to perform analytics and create different events / orders / control or whatever it does. The problem is that automation guy’s are hard to get in with, they are always working at 100% capacity and they do not need extra workload related to a new App to be tested.

At the end these two problems result in Production developing their own solutions for themselves, which is slower, more expensive and not scalable across the production floor and production locations (if more than one).

Question? Now what if the Production needs a state of the art, scalable solutions from the market, that does not want to incur constant integration costs, that wants to manage all the Apps from one environment, which does not want to have high switching costs and try / buy Apps fast and delete them fast if they do not deliver? Here comes MindSphere IIoT Ecosystem ( I will later explain every word, this is not marketing statement.

The potential answer — is MindSphere from Siemens.

First, what is MindSphere?

To make it simple, this is an abstraction layer, that can be introduced on top of the existing IT landscape (next to it, within it etc., actually does not depend on existing IT infrastructure), connected with the Assets at the Production using the required sensors or directly to the existing PLCs, perform EDGE computing if required (e.g. calculate Fourier spectra on the fly) and store this all data in a structured hyper scaler data lake (e.g. AWS, Azure, Google etc.). Thus MindSphere creates a simplified Digital twin of the Industrial automation layer of all its physical assets and brings it to the structured data lake. After this is done, one needs to create and configure user groups, access rights for different groups and enjoy lot’s of different Apps, available from the MindSphere App store (Mendix layer), try them, delete them, buy them. Within the Store each App has a description that explains the integration process, required hardware, price etc. Below one can see a layered architecture which is represented within the MindSphere solution.

Second, next question, that comes into the mind: is the MindSphere is a game change for the Industrial software? The answer is that time will show, since right now it does not have a required scale in terms of apps and number of integrations across the Globe, that would allow it to become the Industrial Marketplace for different industries types, but already today one can say that it is transforming the way industrial software is being built, integrated and managed. It offers a new set of possibilities to exploit business models, which have been widely used in B2C and proven to be more efficient, with good unit economics, while having no switching costs and having very reliable purchase process, like AppStore or Google Play. Finally, we one can see a light at the end of the tunnel for B2B industrial solutions to be scalable with low CAC and acceptable LTW.

Manage your own Apps better from one place. Another interesting application of the MindSphere, despite if there are required Apps or not — is the ability to scale your own Apps across the lines and factories. In my understanding, using the MindSphere for this purpose makes a lot of sense to get more control of the “Apps Zoo” I wrote above. It structures the way Apps are managed, structures the data sources underneath and structures the assets, producing this data (the dream of data scientists). That allows to build own apps focusing on value proposition, while security, connectivity etc. will be managed by the MindSphere. Not bad is not it? In my understanding, that would reduce costs for inhouse development by at least 50–70%. and would give a new breath to in-house acceleration programs, would allow to create perfect sand-boxes for internal and external customers, it can even allow a completely new business models for the corporate apps to exist.

An example: Imagine you have a critical machine at your production — a carousel turning machine for turbine rotors. It’s old and has few vibration sensors and motor parameters (temperature, speed, consumption). But it is crucial for your production. each 1% of capacity improvement makes sense for you. Now one can consider a business model, where equipment data is streamed outside using one of the APIs offered by the MindSphere. Than external developers / data scientists /… can build state of the art models and offer this to the production, that device manager immediately sees the results obtained using the real data and can purchase the best algorithm / app / api etc etc. Without the MindSphere it would be impossible to easily switch the App, since the switching costs are traditionally too high.

Why I said its an Ecosystem. MindSphere, besides it’s abstraction layer for Industrial automation and Industrial Edge has one more layer on top, which is a Mendix solution. That gives a possibility for MindSphere to be an Open platform, meaning others can easily develop for it. It offers a set of things, e.g. Develop apps, Make use of an open environment for development and operations, Get ready-to-use APIs and cloud service, select most fitting your purposes cloud infrastructure: AWS, Azure, Alibaba and Connect with a thriving community of developers and solution partners.

This is available thanks to Mendix, which offers Mendix is a high productivity app platform that enables you to build and continuously improve mobile and web applications at scale. The Mendix Platform is designed to accelerate enterprise app delivery across your entire application development lifecycle, from ideation to deployment and operations.

Thus, MindSphere covers the complete value chain from the equipment at the Automation layer up tp the Apps Store and tools for developers and that is what I call an Ecosystem.


One shall not look for state of the art Apps within the MindSphere at this point of time, it has different solutions and use cases, one can be impressed, one can be not, for instance, I am not that much impressed.

But the game change not in it’s use cases, the game change is in the layer it creates, making it possible to have scalable industrial solutions, changing the purchase process for industrial applications and the way industrial apps are being build.

MindSphere and Mendix platforms:

  • Create a completely new purchase process like an AppStore / GooglePlay;
  • Make it possible to easily scale new apps across lines and factories if the asset structure is more or less the same;
  • Allow to Faster / Cheaper develop your own Apps;
  • Reduce integration efforts for the new apps and thus reduce apps Switching costs;
  • And even allow one to benefit from his/her own developed apps if he/she publishes them in the Apps Store and other companies buy them creating the additional revenue streams.

If this product will be successful and will really become a B2B Industrial marketplace, we will see in future, but the fact, that this is the most state of the art solution in the IIoT Industrial software — for me is already at no doubts. Which Manufacturers it fits better in the current stage — I would think about Process manufacturing first, but the amount of APIs for PLM, ERP, MES etc. is growing so fast, that in my understanding for Discrete manufacturers it can be definitely of value for the predictive asset maintenance (like Process manufacturers), but also to simulate complex production processes.

More one can read here:



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Dr. Alexey Minin

Dr. Alexey Minin

Consultant on Digital Economics, Ecosystems and Digital business models. PhD in AI @ TUM, Honored professor