Enterprise, Lakes, Clouds, and Artificial Intelligence – a Recipe for Magic

You know what’s great about enterprise software?

I don’t have to juggle multiple computers with their own installations—who has time for that?
I won’t lose any work because everything is managed from one central location, so it’s okay if someone spills coffee on a laptop.
Shared files really mean shared files, making collaboration a breeze—no more saving in multiple places!
I can set up a solid file security system since everything is centralized.

And the best part? Enterprise software helps us collect data more effectively, opening the door to metrics, larger studies, and even leveraging artificial intelligence for smarter business decisions.

Lakes and warehouses

When it comes to data, it ultimately has to be stored and for it to be useful, it needs to be structured.

Storage techniques have two main options:

Warehouses: Here, data is organized into a specific structure, making it easy to access and analyze.
Lakes: In this setup, data is stored more freely and structured later, which offers flexibility.

Each storage option has its own strengths and weaknesses.

If two warehouses have different schemas, you can’t just structure them together as-is; just like when merging two different spreadsheets, the data needs to be column cross-matched and sorted first.

Figure 1 – Data Warehouse Structure

In a data lake, the data is all stored together, but to use it, there needs to be a tool that can search through the various file types at once. An example would be SharePoint’s “find” function. It is built to look through Word, Adobe, and Excel files simultaneously and can search through the actual content of the file without necessarily using it.

Figure 2 – Data Lake Structure

Cloud (as a) location and cloud software

Think of the cloud as software running on a network of servers. The setup can vary widely, but the goal is to create a centralized space for enterprise software.
You could consider a LAN-version software as a first step toward cloud software; however a LAN-version is restricted to a handful of individual computers, with individual game installations, connected through wiring.

Figure 3 – Classic LAN party

Just because you install software in the cloud doesn’t make it cloud software. For example, if I run “Worms” in a DOS emulator on a server and give global access, I’ve created a (retro, and awesome) cloud location, but not true cloud software.

Figure 4 – Worms (1995) from Team 17

While cloud software can take advantage of a cloud location to appropriately resource itself, the difference is fundamental to how it is built. Cloud software features coordinated, repeatable functions that can scale and share the workload efficiently.

Imagine transforming “Worms” into cloud software. Now I have an environment whereas many players as the cloud instances can replicate, can join in a single gaming platform, while still experiencing their own perspective of the game. I’m calling it the “Worms Cloud”!

Figure 5 – AI prompt “a cloud of worms”

Machine Learning and Artificial intelligence

What happens when enterprise software, a data lake, and the cloud come together? You get a powerful recipe for machine learning!

In simple terms, machine learning analyzes data to generate statistical trends which can be used to predict outcomes, while artificial intelligence uses trends and outcomes to make predictions. The ‘magic’ of AI lies in its ability to produce results based on vast amounts of historical data.

However, there’s a catch: the accuracy of AI’s outputs depends on the quality of the training data and the parameters it understands. Where data is completely unstructured or the instructions are unclear, it can produce some strange results—just think of those quirky AI-generated hands!

Figure 6 – AI generated hand

The real dream state

Playing ‘Worms cloud’ and generating ‘funky human hands’ sounds like a great weekend after a few drinks, but what can we apply this technology toward in the laboratory?

At one point in the past, I’ve stood in front of an instrument and loudly complained “what is wrong with you?” – what if it answered?

Figure 7 – AI prompt “a laboratory instrument speaking to me”

An instrument speaking might be really odd – but what if it could have produced a report that notified me there was an issue building?

Or what if it called service for itself before it ever broke?

A recipe for magic

To create some real magic, you need a few key ingredients:

Centralized data organized in lakes or warehouses.
A powerful cloud application to quickly process that data into valuable training sets.
Machine learning to turn all that data into meaningful insights.
AI to generate outputs based on those insights.

Unfortunately, the ingredients that are more often found in a laboratory are unique file types, unique applications, and unique technology that aren’t really connected.

While no solution is perfect and no laboratory is the same, we at Thermo Scientific think with the right connected platform, we can leverage more of the rich data sets across more applications. We want to enable scientists in finding statistical significance, training futuristic models, and create amazing “magic”.

To learn more about our connected platform visit the Thermo Scientific Ardia Platform webpage or view our recent Unlock the power of connectivity webinar.

Visit us on LinkedIn: #ArdiaPlatform #Ardia #DataManagement #ThoughtLeadership