Adopting AI - Three Step Process

Imagine if, an AI enabled virtual assistant could understand customer challenges using natural language and provide solutions quickly and effectively on the channel of choice (voice, text, video), delighting the customers

Imagine if, an AI enabled capability, can predict your company’s cash flow, provide propensity score for all the leads and opportunities created by your sales team, etc. which can self-learn, and provide accurate results at a speed far better than any human.

AI is not just for start-ups to take advantage of, today every organization wants to adopt AI, make AI mainstream in their businesses to becoming an AI enabled organization. 

1.  The AI Journey

Unlike the traditional software development, AI journey is a little different, and requires huge amount of compute power, resources, talent, relevant data, etc. to be successful. Therefore, the real success of the AI journey depends totally on keeping the AI journey simple.

The options available for an organization to adopt AI are 1) Solution Specific AI (SaaS Solutions), 2) General Purpose AI (Ready to use Public Cloud AI API’s), 3) Tailored AI (Developing Custom AI Models) which will provide the AI Capabilities to your organization (as shown in figure 1).

(Figure 1: AI Capabilities)

1.1 Solution Specific AI

The Solution Specific AI offered by SaaS (Software-as-a-Service) platforms are, perhaps limited, made to suit a specific need, are prepackaged, provide centrality, which adds value to the organizations. By design, SaaS platforms are commodity and are driven by the configuration data exposed to  the SaaS platforms – this makes the SaaS platforms easier, cheaper to adopt, and less risker to connect with your companies business process as simple as plug and play.

For example, if you look at Microsoft Dynamics 365 AI for Sales, it comes with inbuilt features based on statistical modeling and machine learning to draw insights from the data for both sales users and sales managers (as shown in figure 2).

(Figure 2: Microsoft Dynamics 365 AI for Sales)

These insights are key for the sales users and as well as the sales managers. For the sales users it can help them to identify,

1. What are the key deals they should be working on?
2. What is the relationship and reach within the customer organization like?
3. When was the last interaction or contact with the customer made? 

For the sales managers, it can present them with some key performance analytics and help them uncover trends to drive better performance through their sales organization. Microsoft 365 has a ton of AI driven features such as Microsoft PowerPoint Designer, Microsoft OneNote’s ink to text, and Microsoft Word’s grammar, etc. that help the customers in reducing the time from drawing board to actual solutions.

1.2 General Purpose AI

The General-Purpose AI are Ready to use Public Cloud AI API’s offered by leading public cloud platform providers like Microsoft. These APIs hide the complexities of the AI technology and makes it extremely easy to for the developers to build solutions using them. These APIs offer services such as computer vision, speech, language understanding, etc. and address some of the common user cases across the industries.

To use these APIs, developers do not need to be a mathematical geek or deep learning expert or an AI pundit. These APIs are easy to use, no hardware or installation needed and available in pay-as-you-go pricing model to reduce the risk for the organization. Figure 3 highlights the ready to use APIs available on Microsoft Azure platform.

(Figure 3: Azure Cognitive Services)

1.3 Tailored AI

The Tailored AI is an offering that most of the public cloud platform provide to build Custom AI Models. When the Solution Specific AI and General-Purpose AI doesn’t address the needs of an organization or create an organization specific AI, that are flexible and provide agility. The comparison of the AI options and their advantages are shown in figure 4.

The custom AI models can be created using frameworks like TensorFlow, Keras, CNTK, Scikit-learn, Theano and PyTorch. These frameworks very mature and provide high potential to drive an organization’s AI ambitions. To build custom AI models, organizations need specialized data sciences team, tools, process, and data to train the models.

(Figure 4: AI Options)

Summary

To summarize, many industries have been revolutionized by the widespread adoption of AI and many of us use AI enabled solutions like autonomous vehicles, virtual assistants like Siri, Alexa, etc. AI can add immense value to organizations and help them to transition from insights to action. An organization specific focused AI strategy can help them to leapfrog into the AI worlds.

Banner Image Credit: Joshua Earle