As artificial intelligence (AI) gains momentum across enterprises and industries, we’ve entered an era of Intelligent Automation.
This much more than a simple transfer of tasks from man to machine. The real power of intelligent automation is to fundamentally change traditional ways of operating and transform what people can achieve by weaving systems, data and people together.
Many of the most exciting and disruptive developments are emerging from open-source ecosystems, meaning businesses urgently need a flexible, efficient and secure way to implement best-of-breed AI solutions, whenever and wherever they will deliver maximum value to the business.
The Accenture Institute for High Performance, working with Frontier Economics, examined the considerable economic upside and the obstacles we must overcome to attain full benefits.
We’re seeing growing interest and investment in AI-related technology by Fortune 100 companies in almost every industry. In financial services, for example, robo-advisors, intelligent underwriting and automated agent-based banking services are already in action.
However, the multitude of options notwithstanding, adopters are being confronted with some tough decisions in their journey to enterprise AI, including:
|Cost - Vendor offerings, which are often limited in business application, can be expensive (proof-of-concept work for some applications can cost at least $1 million).|
|Data residence and security - Companies may not be comfortable sharing sensitive data in public or third-party administered clouds, posing challenges for machine learning and data analytics.|
|Customizability and ease of use - Because software-as-a-service end-points can be opaque and inflexible, it can be challenging to customize AI solutions/modules to business needs.|
|Vendor lock-in and technology obsolescence - AI is a rapidly maturing and fast-changing field, and if you tie yourself to a single vendor, you run the risk of technological obsolescence.|
|Future-proofing - The latest AI technology, such as deep learning, is growing in sophistication at a rapid pace, meaning businesses must be consistently ahead of the curve to exploit the potential; that’s not possible when you’re tied to a single vendor that may become legacy at any moment.|
The Accenture Artificial Intelligence Engine (AAIE) has been developed specifically to overcome the challenges companies encounter with vendor-specific Artificial Intelligence options.
Far from tying users to a single AI software provider, AAIE was developed to drive the continuous harvesting and reuse of best-of-breed components.
AAIE follows a transparent, layered architecture whose core capabilities can be used to create recipes for the higher-level services, which can themselves be further trained with data and rules for a particular domain and purpose.
The emphasis is on flexibility. Through an intuitive, graphical and interactive user interface, the engine gives data scientists and Artificial Intelligence application builders enough transparency to choose the sets of components they need for a particular task.
Many enterprise Artificial Intelligence applications, typically closed-domain and complex, present challenges, including insufficient data for training, lack of trained experts and inadequate flexibility.
AAIE addresses these issues head-on, providing:
Faster time-to-build through a rich technology and domain library of harvested components.
Data science know-how through an intelligent user interface studio, which enables “drag and drop” and auto-validation capabilities.
Greater flexibility through a scalable and modular micro-services-oriented architecture, access to the best technology solutions through an open, agnostic innovation model and “open access” source code enabling higher customizability.