eWEEK content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.
The funny thing about artificial intelligence and AI derivatives such as machine learning, deep learning, and now generative AI is that they’ve been around for so long. While those entering the IT market have a renewed understanding of AI, its core mechanisms pretty much remain the same.
The cost and the ease of building very complex and valuable AI models are what has changed. The use of cloud computing ignited the rapid rise of AI systems. As businesses expand their investments in cloud technology innovations, AI can provide still more business value.
Most enterprises leverage AI systems as monolithic systems. They receive training data, learn from it, and then leverage those models to carry out some predefined purpose. For example, determining the credit risks within a loan processing system, automating a supply chain to minimize inventory, or picking stocks for an investment firm.
However, the innovations available within these standalone systems limit the value of these monolithic AI systems. This limitation created the drive to integrate AI technology with other core business systems and other AI systems to add value.
When done correctly, the end state is 1+1=3. This will also increase an AI system’s ROI, which can be costly to develop and maintain.
Also see: What is Artificial Intelligence
AI Orchestration and Integrating AI Systems
Integrating AI systems with many other business systems (either loosely or tightly coupled) is nothing new. AI systems are featureless, at the end of the day. While they can carry out some impressive functions, they are not business applications unto themselves. They must be bound to business applications to create the business value needed from AI.
The trend in the past was to tightly couple AI systems to applications and application data. The AI systems were either a component of the business system that runs in the same space and on the same platform or the systems that are accessed through synchronous (blocking) APIs that can programmatically interact with the AI systems.
The tight coupling of AI systems to business systems was all about choosing the path of least resistance. While it “worked,” it limited the value AI systems could provide to many other business systems.
Lightweight orchestration services can deal with the unique requirements of AI systems and provide loosely coupled access to any number of AI systems. They leverage those systems using no-code or low-code development, where orchestration services create solutions by using configuration rather than deep development. Loose coupling vs. tight coupling allows access and orchestration to be set up to leverage AI systems in ways that can drive more business value from those systems.
Orchestrations are neither new nor innovative, whether they run within the same cloud, across clouds, or across internal systems. It’s a new application of an existing innovation that provides the ability for orchestration layers to deal with AI system interaction in ways that make the AI systems much more effective when leveraged together (1+1=3).
For example, let’s say we have an AI system to pick stocks that will increase in value, an AI system to support risk analytics for specific companies, and an AI system to spot trends in social media before they become trends. While each system has clear value by itself, the ability to create orchestrations that can evaluate specific business problems or proposals through the combination of all or some of the AI systems has much more value.
Referring to the above example, we can use a list of stocks created by the stock-picking AI system and then pass that information off to the risk analytics system. That system can cull through the data and assign risk rankings based on its training data and the resulting knowledge model.
Then, we take the risk-ordered list of good stocks to our social media AI-powered analytics system that can assign good or bad trends to specific products that those companies produce. This combined application of AI systems can further determine if the market will likely support that company.
You get the idea. Follow the patterns of process integration, RPA, data integration, and other systems that operate above more primitive systems, which allow more value to be extracted. This is about supporting automation between these systems to find useful orchestrations that can be built as an ad-hoc solution, or as something more permanent.
Also see: The Future of Artificial Intelligence
Full Potential of AI Orchestration
Of course, our simple example does not illustrate the full potential of AI orchestration or the number of potential business applications. Other examples include:
- Orchestration of 12 different AI systems and 110 databases to support supply chain integration that allows decision-making with near-perfect information and understanding.
- The ability to create an investment system like the one defined in our example that picks stocks with an 89 percent success rate using dozens of AI systems and internal and external data sources.
- A system that ranks the best potential employee candidates using external and internal data, and 9 AI systems that can determine which candidates are most likely to be successful in a specific role.
AI on its own has some value. Integrating AI with other AI systems and other data sources can produce much more value. This is the point where companies can take AI to value levels that have not been reached. This also provides a platform for experimenting with the power of AI, which can, in turn, find innovative values that most businesses have yet to realize or even conceptualize.
Also see: The History of Artificial Intelligence
Two Types of AI Orchestration
So, what does an AI orchestration tool look like? There are two general types:
- The first type includes workflow and orchestration tools that natively support AI integration.
- The second type of tool requires you to extend its capabilities to interface with the AI system and deal with AI data to be generated and consumed (e.g., training data).
Some may even leverage AI for their workflow or orchestration capabilities, but that’s not on the critical path. The critical path is to leverage a tool that can integrate your source and target databases and AI systems, preferably using a low-code or no-code orchestration engine that you can easily and quickly setup. The platform you leverage (e.g., public cloud), the enabling technology, and other factors will determine the “right” tool AI orchestration tool to leverage.
Again, this type of AI is not new or science fiction. It’s simply an approach that leverages existing technologies to bring more value to the business. That said, most businesses don’t yet understand what AI orchestration is, and fewer recognize its game-changing value for businesses. It’s time to do some homework.
Also see: How AI is Altering Software Development with AI-Augmentation