Companies need to take the AI brain “out of the jar”; using application networks that enable AI to plug in and out of any data source or capability.
Everyone, including MuleSoft, is talking about artificial intelligence (AI). The technology might still be in its nascent stages, but we’re already getting a glimpse of its promise. The opportunity to automate manual tasks and gain new insights to support business decision-making could have a huge impact on organizations; so much so that IDC forecasts businesses will be spending $57.6 billion on cognitive and AI systems by 2021.
However, as with any technological innovation, there are challenges to overcome before we will start to see AI’s full potential. While most talk about the desired benefits of deriving insights from massive data sets or automating business processes, most overlook how difficult it is to make the connections needed in the enterprise to achieve those outcomes. Siloed data stores and a lack of connectivity between enterprise applications severely restrict AI’s current ability to influence the digital ecosystem around it, rendering it little more than a rather expensive brain in a jar.
The trouble with AI
The C-suite might generally be on board with the need for AI, but in most cases they’re still trying to learn how it can be leveraged strategically. Where it’s being used perhaps most often, and where we’ll likely see most early use cases, is in analyzing large sets of data and identifying patterns to generate new business insight. It’s all about predicting behavior, based on the outcomes of thousands of other similar situations that have occurred in the past—a task virtually impossible for a human to do manually given the quantity of data involved.
In addition to the volume of data, another challenge lies in the structure and quality of the data. Organizations are often restricted by the fact that data is locked up in siloed systems and applications. As a result, getting that data into an AI engine to start revealing insight can be a major problem and limit AI’s potential to offer competitive insight. To get the most value out of AI, it’s not a case of knowing the right questions to ask but having the ability to connect AI engines to the right sources of data.
Automating the future
Business automation is one of the major areas that AI is set to disrupt. For example, JPMorgan has created its own AI engine, LOXM, which uses the lessons learned during billions of past trades to quickly execute deals that secure the best price. This type of automation can translate to other industries too; retailers will often spend hours trawling through sales and footfall records to identify staffing requirements and allocate resources to their store network. It’s currently a laborious and inefficient task. However, if the retailer set AI to work on that same data, the manual process could be replaced by an automated system. The AI engine crunches the data, spots where staff is needed most, and then updates the HR systems to schedule shifts.
Additionally, automating these cumbersome processes will free up staff to work on higher-value activities that drive greater benefit for the business and improve employee satisfaction.
However, the AI brain needs to be able to connect with enterprise systems if it wants to automate tasks. To illustrate this with an example, one company I’ve encountered has been working to create an AI-powered chatbot to help with customer support. However, they hit a brick wall because even a task as simple as updating a customer’s details required AI to be integrated with seven or eight different systems. The truth is, if you don’t have this connectivity fabric in place, AI is capable of thinking, but not doing.
Making the brain in a jar intelligent
These challenges largely stem from a misunderstanding that AI will just plug into an enterprise like a new brain. Unfortunately, some AI vendors have further fueled unrealistic expectations by trumpeting various big-name projects, without clarifying that these capabilities can’t be reused to achieve different outcomes. This will only add to buyer dissatisfaction and technical debt for those that leap straight in without first laying the foundations for AI.
To do so, organizations first need to become more composable, building a connected nervous system called an application network, which enables AI to plug in and out of any data source or capability that can provide or consume the intelligence it creates. The point-to-point integrations of the past won’t be practicable in the AI-world, where things can change in an instant. Instead, organizations need a much more fluid approach that allows them to decouple very complex systems and turn their technology components into flexible pieces.
This can best be achieved with an API strategy, which enables organizations to easily connect together any application, data source or device into a central nervous system of sorts, where data can freely flow. The central nervous system, or application network, is how the ‘brain’ of AI can plug into a business’ digital ecosystem to consume its data and then provide valuable insights and actions. In the case of the customer service chatbot I mentioned earlier, with an application network, the business could enable AI to access any system where customer information is stored and then instantly update it with their new address.
The future’s bright
While we’re still some way off mainstream adoption of AI, we’re already seeing impressive results. For example, one insurance company I spoke with is using AI to discern which of its customers are most likely to renew their premiums, and then automatically contact them via SMS. The firm saw a 60% engagement rate and 30% renewal rate from that text message bot campaign alone.
If connectivity challenges are overcome, the potential for AI to uncover valuable new insights will change the speed of business, while automating repetitive tasks will to free up organizations to focus on more value-add activities.
This article was originally published on CBRonline.com.
