Digital Twins for Organisations.

Imagine we’ve just read the latest Gartner report citing the impending ubiquity of Digital Twins and have decided to build a start-up around the concept. We’ve defined “Digital Twin” as:

“integrated stream(s) of data that express the current and past state(s) of a complex system in at least three dimensions (E.g. time, location, activity / cargo type, capacity, traffic / crop type, weather, yield etc.) such that the user’s understanding of that system is improved.” 

There are other definitions out there, including this great write up by IBM you can find here, but we prefer ours since it doesn’t limit us to physical space.

We might start with the customer and ask why they find Digital Twins valuable. The response would likely include some combination of the following themes:

• Current State - What is the current state of my system?
• E.g. How many centrifuges are currently spinning?
• Historical State - How does the current state of my system compare to historical states?
• E.g. Is it normal for 25% of my fleet to be off-line on a public holiday?
• Simulation - If I adjust the twin’s inputs (temp, range, mass, throughput, visibility, price, latency, etc), how will it respond?
• E.g. If I had 30% off-line, could I keep up with demand?

In some cases the ‘twin’ might focus on different elements of the above, depending on the nature of the system (the data streams required to understand an industrial system are vastly different from those relevant to a transportation network), with different levels of granularity being useful for different purposes (relevant perspective on a twin of an engine will differ greatly from the ideal perspective of a national power grid). 

Whatever the data type or perspective, we elect to store in a Lakehouse architecture, which ensures scalable storage with stream processors integrating data into a flexible data model, and AI-enhanced interfaces contributing to a concise and intuitive user experience across user types. All of these services are available to one another as microservices within either a single or multi-tenanted account. Our goal is to make this model/‘twin’ as relevant as possible to decision makers, be they in the back office or on the front lines, and a blend of real-time stream processing and Lakehouse for ad hoc analyses allows us to do that. We believe that these types of models will improve decisions and save lives. 

Therefore, rather than focus on a specific existing market for bespoke Digital Twins in a domain like industrial management, manufacturing, robotics, and logistics, our product will be applicable to any mission-essential system, be it; IT service orchestration, fleet/asset management, investment management, law enforcement or defence. We achieve this generalisability through the introduction of a prescriptive, event-based data model called SMAQ at the point of integration. The assumptions baked into the SMAQ data model allow SMAQstack to append contextual enrichment data from common sources (weather, traffic, geodata, social statistics, etc.), merge data from multiple sources for the generation of insights and extend the ‘twin’ concept to new domains at a reasonable price point. 

We can now quickly understand the state of the system, whether that state is normal, and what may happen if a specific event occurs. Infrastructure as Code ensures that each system is backed by its own isolated cloud environment, and capable of scaling up to manage any level of complexity or data engineering load. 

Among other products, we at bolster are using the SMAQstack concept to build this engine, such that it is a useful facsimile of real mission critical systems, and we are looking for beta customers to work with us to model their complex system using the SMAQ methodology and platform.

What would you do if you had Digital Twin(s) of your: Sales team? Warehouse? Supply Chain? Blue Forces?

Sign up for the newsletter to stay across how bolster can create a digital twin for your organization.