In a recent podcast interview, Microsoft CEO Satya Nadella suggested somewhat offhandedly that generative AI—specifically AI agents—would soon render enterprise SaaS applications obsolete. While Nadella may have been trolling, his comments certainly sparked intense debate and conversation. But what does his pronouncement even mean? 

Given the recent ability of multimodal LLMs to take control of one’s computer and perform actions using the application’s Graphical User Interface (GUI)—Anthropic Claude with computer use is just one example—Nadella is envisioning a world where LLMs are unleashed to update applications with no need for humans to touch the keyboard. 

In fact, the Microsoft CEO goes even further, imagining a new landscape where GUIs are no longer required. As impressive as it is for AI agents to autonomously control business applications via a GUI, GUIs were created for humans. AI agents don’t require a GUI, but can instead interact with databases directly. Enterprise SaaS applications, Nadella says, are really nothing more than a GUI on top of CRUD (Create, Read, Update, Delete) databases. Correspondingly, if AI agents don’t need GUIs, and SaaS applications are essentially GUIs, why do we need SaaS applications in the approaching agentic world? 

The Bright Side of an Agentic Future 

As someone who spent decades working with ERP systems, I’m personally enthusiastic about a future where business professionals—like procurement specialists, accountants, and HR professionals—will no longer need to learn the quirks of a complicated software system and spend much of their time doing data entry.  

How great will it be when an employee can tell her AI assistant “I need to reorder 500 of these widgets” and the AI can execute all the tedious data entry tasks to update the appropriate backend systems and shepherd the approval workflows? Now the requester can go about her real work, while the procurement specialist can concentrate on truly value-added tasks, like negotiating a better price for the widgets, probably also aided by his own specialized AI assistant. 

Back to Reality – The Challenges 

But of course this “glorious future”—to borrow a phrase from Sam Altman’s recent blog post on the advent of super intelligence, is not yet here and there are hurdles that stand in its way. In an insightful essay, my former NASA colleague Randy Sparkman points out that businesses run on many entrenched “deterministic” legacy applications (some SaaS and some on-prem) and these applications are not all pre-wired to be controlled by “probabilistic” generative AI agents. Even if they were, AI agents may find that the rigid business constraints of the legacy systems hamper their objectives. 

Beyond these technical hurdles, there are still outstanding issues of trust and control when it comes to handing off critical enterprise functions to probabilistic AI.  

What Does the Agentic Future Mean for CCaaS? 

Contact centers are still largely run by humans, for humans. However, now that LLMs have made it technically feasible and affordable for computers to understand spoken and written human language, CCaaS vendors are helping contact centers quickly install powerful self-service capabilities that successfully handle significant volumes of repetitive interactions. Yet beyond these self-service capabilities, the largest portion of the CCaaS product portfolios consist of software designed for humans, be they agents or the supervisions and managers responsible for overall performance of the contact center.  

In fact, much of CCaaS pricing is still based on the “per seat” model, where customers pay for the number of human employees with access to the software. If the current generative AI momentum continues, and agentic AI becomes capable of handling an increasing portion of tedious backend tasks–from knowledge retrieval to source system updates–where will that leave CCaaS vendors? 

How CCaaS and Humans Can Evolve and Thrive in the Agentic Era 

Despite the obvious threats to SaaS from the impending agentic wave, CCaaS solution providers may be in a more enviable position than other SaaS vendors. Their software provides a GUI through which humans can interact with customers and generally involves connecting to backend SaaS systems for data such as customer info, purchase history, product information, trouble tickets, etc. AI agents are more likely to disrupt those backend systems. If AI agents become the “hands” that do the data entry, humans can still be the brains and the voices that talk or text with customers, especially for higher value interactions.  

The best CCaaS solutions offer transcription for all voice calls and logging of all voice and text interactions, which forms the basis of a powerful analytical component to provide agents and supervisors with insights into why customers are calling and into how well agents are handling these calls. These insights are just as valuable to AI agents as they are to humans. 

It’s really up to the brand to determine how much of the communication with the customer they wish to automate via LLM-powered voice or chatbots for “self-service” and what types of interactions they wish to preserve for human agents to deal with. In the example used earlier of AI agents removing the data entry burden from the procurement process, AI agents can gradually remove time-stealing data entry burdens from human call center staff. Just one example is the manual creation of call summaries, which many companies have already handed off to LLMs, thanks to their CCaaS solutions.  

The rise of agentic systems opens new avenues for innovation in CCaaS. Advanced analytics, automation, integrations, observability and safety will be critical in enabling both humans and AI to detect issues, analyze patterns, and promptly execute resolutions in a trusted manner. Vendors can leverage these technologies to power agentic systems that enable brands to relieve burdens from human agents–such as updating backend systems and searching for answers–or completely automate interactions as they see fit. 

The Road Ahead 

It remains to be seen whether Satya Nadella’s prediction for the demise of SaaS comes true. But either way, it’s clear that we’re not in Kansas anymore. Much like Dorothy stepping into Oz, we find ourselves navigating an environment where familiar landscapes have been upended, replaced with possibilities both exhilarating and daunting. At Opus Research, we’re excited to explore this new landscape.  

Categories: Articles