OutSystems new artificial intelligence capabilities give you the power to create smart experiences and automation with chatbots and text & speech analysis. These allow organizations to harness automation to create self-service portals, respond to text and voice queries, improve customer service, and much more. You will find a number of AI high-level components and connectors available in the OutSystems Forge for you to use and incorporate machine learning features in your apps.

How Did we Get Here?

The modern world changed when electricity took over. Everything from transportation to healthcare was revolutionized, the world was more accessible, and a whole new array of opportunities opened up in front of humankind. As we progressed further through time, we created computers and software followed soon. Today, companies across the globe are either entirely driven by or driven using software. Continuous innovation in the field of technology ensured that we were no longer happy about just being software-driven.

We wanted better, smarter applications. With the rise of artificial intelligence and its role in software in the current landscape, everyone wants to have software applications with AI capabilities. According to Forrester, three out of four companies want to implement AI features in their products in the next 6-12 months. 87% of those implementations will target efficiency gains and automation, while 55% will aim for revenue increases alongside personalized experiences. However, two noticeable barriers to adoption of AI features in business consist of a lack of talent (people who know how to work with these technologies) and a lack of understanding of the possibilities and use cases. This is where OutSystems comes in, enabling you to use low-code to create applications that have AI capabilities.

Surely, AI Won’t Affect Our Line of Business...

Soon, the difference between applications with AI and regular applications will be the same as smartphones and “normal” mobile phones today. The fact that today we don’t even know how to name those phones should indicate what fate awaits software applications.

Studies show investment in AI is increasing, and it’s not showing signs of decreasing in the future due to the many successful use cases that have become public. Artificial intelligence is, currently, the stuff of modern success stories that have become legendary. Analysts like Gartner are estimating that 2020 will be the year that applications leveraging AI capabilities will take over and that the world’s largest companies have plans to develop intelligent smart apps.

Company investments are usually made with one or more of these goals in mind: increasing revenue, reducing costs, and avoiding risks. As such, applications that take advantage of AI capabilities successfully fit the company’s purposes.

How Can My Apps Benefit From AI Capabilities?

Incorporating AI in your business ventures can result in amassing significant economic value. You can be on the lookout for increasing your company’s revenue. Your customer-facing applications can now have chatbots with business logic that help your customers, advising them on the best course of action or clarifying their doubts. All of this helps your company generate better leads and ultimately creates conversions while still keeping your customers highly engaged and satisfied.

AI can help reduce costs and increase productivity by removing existing points of friction, such as the process of information submission. Let us consider the scenario of an insurance company with an app that its customers use for insurance claims. Instead of the inconvenience of filling out forms, customers can take a photo of their cars after the incident. This photo, along with other collected information like location, time, and weather conditions (that can be permissions to be enabled in the app,) can automatically start their insurance claim process.

When you think about it carefully, you realize that in the back office of the insurance company, the processes would follow the same flow; however, the process could be delayed by several manual decisions and actions. A machine learning model could be trained to automate these decisions to a certain degree of confidence, which leaves the reduced number of remaining cases to still be manually decided.

With a well-trained chatbot, customers can successfully make and follow the insurance claim and pose questions to ensure that everything is being taken care of. This, in turn, removes pressure from the first-line call center operators.

When it comes to reducing risks, AI allows detection of anomalies and helps build fraud detection mechanisms in systems that support millions of transactions. For a person, the task is similar to looking for a needle in a haystack. However, as difficult as it is, the seemingly insignificant needle can have a significant impact on results.

Okay, I’m Sold. How Can I Use AI in The Apps I Build With Low-Code?

We’re glad you asked. There isn’t yet a massive implementation of AI in applications. As a result, organizations that don’t have dedicated teams and tools for AI are struggling. The main difficulties can be narrowed down to three main topics: understanding what can be achieved with AI, access to data, and access to talent.

First, let’s clear up what can be done with AI. It is of utmost importance that there is enough knowledge on the actual capabilities and what problems can be solved with AI and machine learning. Holding on to a science-fiction-inspired illusion may well cause an AI initiative to fail from the very beginning. We are currently in the digital transformation era, but a new one is rising: the AI transformation. Like the playbooks on digital transformation, an AI transformation playbook, too, has been created to help organizations venture down this new road.

Accessing data can be a significant problem if the application that will use AI requires specific business data. However, a lot of AI capabilities like text, speech, and image analysis don’t need particular data. We can train these models with referenced datasets that are already both widely used and shared globally.

In cases where models rely on specific business data for training, the concern over access to quality and quantity of particular data comes into play. It's intuitive that if insufficient and flawed data is used to train a machine learning model, it lowers its capability to perform well.

If your data is in the OutSystems platform, then it’s already accessible. Even if it is in another database, there is always the possibility to access that data from OutSystems with an extension.

AI and machine learning are relatively new, and there isn’t much available talent in the market. We want customers to obtain value from their business data and new AI-driven experiences with little to no data science knowledge.

As such, OutSystems has maintained the same vision from software development, using low-code to remove complexity from the equation.

What Does OutSystems Provide for AI Capabilities in Enterprise Apps?

We are making it easy to adopt and integrate powerful AI capabilities in your apps. We’re providing rich components that you can use with our low-code platform to minimize the amount of glue code. They also reduce the complexity of implementing a complete use case from end-to-end. This opens up the implementation of top use cases of conversational interfaces and text and speech analysis so that it is accessible to everyone.

Microsoft has started to provide AI services in Azure. In tandem with that, we have been creating connectors for those services to be available in the OutSystems Forge so that everyone can use them. These connectors cover a wide range of functionalities like speech, text analysis, image recognition, machine learning, and also multiple providers. You can learn more about these connectors in the OutSystems Evaluation Guide.

The key difference with this new release is that we’re moving now providing higher-level components and not just connectors. We built specific building blocks to help you create applications with AI capabilities. Components are easy-to-use, drag-and-drop elements that satisfy one of these AI functionalities: cognitive services, automated decision-making, and conversational experience.

Cognitive services allow the use of AI in applications for unstructured data use cases such as speech, text analysis, and image recognition. The goal for these components is to help retrieve relevant information that exists but can sometimes be difficult to find. Take the example of sentiment detection in the OutSystems.AI Language Analysis component. In an engagement with a customer, there is more to retrieve that just the fact that it’s raining outside. Ideally, it would be possible to ascertain if the rain is good news or bad news for the customer (or if they’re completely ambivalent to the weather.)

Let's consider the example of improving customer satisfaction for a specific company. The company provides its services to the customer. At the end of the service, an automatic feedback call is recorded.

With our components, it’s possible to transcribe the customer feedback to text, analyze it by capturing the key phrases and detect the general sentiment of the feedback. If the sentiment seems negative, it’s possible to automatically create a ticket for a customer-care agent to follow-up with that customer.

 

OutSystems.AI Language Analysis transcribes the audio, detects the keywords and the sentiment behind it.

Components for automated decision-making helps customers use data in OutSystems effortlessly by adding high-value predictive capabilities and process automation. Based on previous decisions stored in the system, it's possible to train a machine learning model to make those decisions with confidence. Threshold decisions can thus be automated, and suggestions can be shown with a degree of confidence as well. The models are continually being trained, so the more they work, the more accurate the results.

Conversational experiences are the new paradigm of interaction between customers and products (or services.) The rise of virtual private assistants means that these conversational experiences will be the standard soon.

One of the most visible components of conversational experiences is a chatbot. You can use a chatbot in customer-facing application with roles from generating qualified leads to actually making a conversion or even helping in a self-service portal. They can be used in internal apps to onboard people answering the simple everyday questions. Chatbots can also answer quick FAQs and also enable people in their new roles by providing information and access to files that they may need to do their job correctly.

 
Chatbot trained to serve customers in the customer-facing portal by answering their questions and fulfilling their requests.

 

 
Chatbot trained to help the technicians that are working in the field and do not necessarily have all the required information to troubleshoot all kinds of issues and problems.

 

That’s Interesting. Is There More to Look Forward to?

You can find the AI components in the OutSystems Forge. We plan to keep on evolving them, depending on both the needs detected and the feedback received. Take a look around, use the components, go through our documentation, and let us know what you think about them! As mentioned before, we also have an industry-leading set of connectors to the best services in the market right now. Don't forget: what seems to be currently a "nice-to-have" feature will very soon be a "must-have," and this is an excellent step forward for your applications and organization.