SuccessKPI’s Roadmap to Invisibly Integrated AI

As we know, artificial intelligence (AI) did not begin in 2023 with the public launch of ChatGPT, despite its meteoric rise and mass awareness.  AI has been with us far longer than many people realize.   

Oxford dubs AI as: “The capacity of computers or other machines to exhibit or simulate intelligent behavior.” 

This is an automated process. Artificially intelligent machines can remember behavior patterns and adapt their responses to conform to those behaviors or encourage changes to them.  Generative AI has taken this concept to a whole new level where AI can now reason, get creative, and have long memory and context.  

Market Approaches

SuccessKPI has been a leader in the development of AI for the CCaaS market which, of course, impacts a broad range of professionals, from customer service/customer experience (CX) to marketing, sales, business development and anyone in an enterprise that wants to better understand and reach their customers.  We were the first in the industry to publish our AI strategy and roadmap for all to see.  

SuccessKPI’s approach to AI strategy includes: 

1. Use case-driven: enabling a distinct large language model (LLM) for each customer use case.   
2. “Invisibly integrated AI,” in which there is no knowledge of LLM or any of the technical understanding of AI needed; the AI works seamlessly within the fabric of the core product just as an extension of the product.  This approach is the subject of this article. 

Why Invisibly Integrated AI? 

SuccessKPI started down a path several years ago of examining our solution set and determining our AI strategy. We did extensive research across markets and different industries. We detected a clear pattern around a big data science approach that pulls in the data, applies the large language model, trains the model, cleans the data, rinses and repeats. 

This was very sophisticated and because of that, only a data scientist could work with those products and capabilities. Then there was another group of players who were using AI for solving specific problems. That is a use case-driven approach. 

We also noticed another group of players who were applying AI in a very subtle way that is not obvious to the user. Their AI was under the hood, helping customers perform better and more efficiently without AI knowledge or deploying expensive resources. That really resonated with us. AI does not have to be a complex, intimidating thing. It should be baked into solutions, helping different customer personas function better as a seemingly natural extension of themselves.   

This is how we arrived at SuccessKPI’s invisibly integrated AI strategy. 

The Invisible Building Blocks of Invisible AI 

This concept of invisibly integrated AI is not entirely new.  As with most things, it is built on previous innovations.  

AI was conceived way back in the 1930s by British mathematician Alan Turing, who formulated the famous “Turing Test,” in 1950, a method to evaluate a machine’s ability to exhibit intelligent behavior equivalent to, or indistinguishable from, that of a human, which has yet to be achieved.  

As long ago as the late 1970s, my father, who was a professor, wrote a paper about AI at his university in India. He described the principles and potential applications. But instead of reacting to this as a revolutionary thought, no one cared. It seemed too much like an unattainable fantasy, like the sentient supercomputer Hal from the movie 2001: A Space Odyssey.  This period was called the AI Winter. 

So much has happened since then. Siri, the iconic voice of Apple’s iPhone, may be the first application that entered AI into mass consciousness. Open AI’s ChatGPT took our public awareness of AI to a whole new level. 

But before these came predictive analytics. We used to predict what is the best channel to contact a customer at what time of the day with the greatest probability of reaching that customer. It was, and still is, very effective. Gartner and other pundits thought the ultimate evolution of analytics was predictive analytics.  They didn’t see the practical applications of AI on the horizon. 

Common Examples of Unrecognized AI 

Look at Zoom, the popular UCaaS/meeting application. It performs call summarization, transcription and lists action items for participants, among other things. These features work automatically.  They are part of an application experience.  

Staying in the same product category, the new version of Cisco WebEx has a new feature called “Catch me up” that updates participants on what they’ve missed if they join a meeting late. This call summary reduces friction in the product by avoiding all the delays and distractions caused by questions and explanations on what was missed.  

Other types of examples include: 

Maps and Navigation – Google or Apple Maps know where to go and how to best get there.  This is AI functioning on top of satellite-based GPS to give users a much more enhanced experience. Using machine learning, the application has been taught to understand and identify changes in traffic flow so that it can recommend a route that avoids roadblocks and congestion. 

Facial Detection and Recognition – Using virtual filters on our faces when taking pictures and face ID for unlocking our phones are two examples of artificial intelligence that are now part of our daily lives.  

Social Media – Social media applications use AI to monitor content, suggest connections, and serve advertisements to targeted users to ensure people stay invested and “plugged in.”  AI algorithms can spot and swiftly take down problematic posts that violate terms and conditions through keyword identification and visual image recognition. Social media AI also has the ability to understand the content that is relevant to users and suggests similar content to them. 

How Do We Know AI is Working when it’s Invisible to Us? 

Similar to the earlier Zoom and Cisco examples, SuccessKPI leverages invisible AI to assist our users to be better and more efficient.  Personas such as contact center agents, supervisors and evaluators are trying to do many different tasks. SuccessKPI’s platform helps make their jobs frictionless as they go through their daily routines. 

What took them two days is now done in two minutes. It’s that transformative.   

Our solution, for instance, would detect if an agent gets hung up on a certain topic or term.  It would prompt them with a real-time assistive suggestion that could keep them functioning fluidly.   

The AI would know what to suggest to the agent when they’re stuck because of their history when dealing with this particular type of situation.  It knows the context of the conversation with the customer, and it anticipates the agent’s need, suggesting the relevant information because of all the other cases around this particular topic.  It has the knowledge of that company-specific domain. It has insight into thousands of conversations that hundreds of agents are having every day.  

SuccessKPI has a speech product that listens to thousands of calls and identifies emerging patterns around specific topics. People are talking about product problem X or about cancellation or perhaps payment related issues.  When the agent goes to query or type something around that topic, it’s already there as a recommendation.  As these experiences build, the agent can save or deselect certain AI prompts so they become more like a reviewer controlling and editing the content that works rather than having to create something from scratch. 

Another wish-list capability that has manifested into reality recently is this concept of agentic AI. This is a virtual agent that can tackle a specific task or run its own workflow. We see a world coming where your virtual agents can react to changes in call topic patterns and adjust in hours to what now takes months or quarters to program the IVR or upskill human agents. The agentic AI workers can discover those changes in call patterns. They can spin off agentic workflows around those changed or new intents and then immediately serve them. 

There is also now an opportunity through AI advances to reexamine the daily routines of the agents, supervisors, evaluators, speech scientists and other personas who play a role in the CX.  All their work journeys need to be analyzed under the auspices of generative AI as an operating system (OS). Consider how one would design this application from scratch and rethink the whole journey given that the OS is a lot more intelligent, not just a powerful processing machine.  

Completely reimagining those workflows could streamline, redefine, and automate many of the steps and create new functions delivering new value for the business. It’s a redefinition of user experiences based on GenAI as a platform. The problems that need solving are not going to disappear but how they are solved will completely change. 

How Can CX Professionals Adapt? 

Yes, this means that an unknown number of agent jobs will be replaced with AI-powered virtual assistants. But that cycle of innovation and recalibration is nothing new.  Throughout human history, the labor market has always rebalanced itself based on advances in technology as well as skills supply and demand.  

There was a time when stagecoach drivers faced career extinction as automobiles became the de facto mode of transportation. As they went out of business, some found new roles.  Today in the contact center, we see a similar winnowing process that favors those who evolve with the technology and advance into roles that leverage their skills in more impactful ways. 

We evolved into a species where we learned a lot of things from our experiences due to our intelligence. Humans have never been the strongest or the fastest, but we developed to be the smartest, learning from our past experiences. For example, most IT professionals in India 20-30 years ago were software coders.  When confronted with the rapid advances in technology and the talent competition in the US and China, Indians concluded that they needed to evolve to handle more complex roles, to create more value, get into the design architecture and become full problem solvers rather than just handling one link in that long value chain. 

The global CCaaS industry today is no different. We need to evolve into that value chain and do things which are more complex, more valuable, that require more specialized skills and intelligence.   

In summary, we need to stay ahead of the machine. 

Now all the CCaaS players are trying to formulate an AI-driven story for their company, customers and product. This is an authentic, industrywide reengineering of brands. But it requires that evolution within the human workforce to adapt to the new technology to make it real. 

SuccessKPI, by contrast, has already been working with many customers, demonstrating our AI capabilities infused into one, cohesive platform so they can experience it directly today, not in 2025 or 2030, but right now.   

The Quantum Leap: Where is it All Going?  

Going forward, we are headed for uncharted waters. The gap between a human-led CX and an agentic AI-driven experience will increasingly narrow in the coming years to the point where it will close and they become indistinguishable, e.g. the “Turing Test.” 

It could arrive 10 years or 25 years into the future or much faster than we can imagine, like ChatGPT. When we hit that mark, the AI will so completely merge into the CX experience that it becomes invisible. At that point, there will essentially no longer be a need for a contact center. Think about it: CCaaS is a way to efficiently manage the routing of customers to agents. If a brand can create a million instances of an agent, it no longer needs routing. 

This is similar, but not to be confused or conflated with, the concept of the singularity—the physical merger of humans and artificial beings to where we evolve into something completely new. 

In our industry, the new agentic AI will be very human-like in behavior but will have the intellect of the entire brand/market/industry because each agentic instance will have the ability to access all the information that is available. This is a world without software or hardware products. Companies will build unique customer experiences around AI for every customer journey. 

Imagine: A customer of a particular fashion brand visiting that brand or calling up their agentic AI brand ambassador who guides them through their journey to help them find, select, and try on new clothes, shepherding them all the way to final purchase.  There’s no need to go to a contact center because that AI guide is the entire customer experience around that brand. 

Many vendors will participate in that “invisibly integrated AI” future—companies like Adobe from the marketing side, Zoom from the UCaaS market, Microsoft from their cloud-based desktop position, Salesforce from the CRM end.  All these adjacent players and others like them will work to create holistic AI-guided customer experiences.  If anyone can spin up a million agents to help a million customers at any moment, scaling resources is no longer a problem.  

SuccessKPI, for our part, is well positioned to be one of those experience creators for two reasons: 

1. Being cloud-native, we do not have that burden of legacy older infrastructure, building IVRs, routing to agents etc. 
2. As a Workforce Experience Management (WEM) leader, we are entrenched at the center of the most advanced CX as it evolves toward that future state. We have all the data and the knowledge of how people interact from both the agent side and customer side. Our established role is to make agents better so their interactions with customers are better. We are already in the heart (and brain) of an ever more intelligent CX experience.