Why businesses are testing conversational AI beyond CRM

Customer relationship management (CRM) tools have long served as a source of business truth. If the data is clean and the fields are up-to-date, they can tell an executive what’s in the pipeline, what’s likely to close, and what the revenue will be next quarter. And more often than not, entire projections, compensation plans, and boardroom discussions are based on the assumption that what’s inside the system reflects reality.

But anyone who has been involved in a recent review of the agreement knows how fragile this case can be.

The moment an agreement actually changes rarely appears as a field update. It happens behind the scenes when events change before systems can catch up, like a buyer hesitating on a call even though the deal still looks healthy in CRM.

This disconnect is why a growing number of businesses are experimenting conversational AInot as a replacement for CRM, but to see what CRMs usually miss.

The problem with CRM

CRMs are great at structure. They record names, dates, deal sizes and business stages consistently. However, what they struggle to capture is the context—the subtle signals that shape outcomes in real time.

“CRMs are systems of record, making them only as stable as the manual data people feed them,” says Carson Hostetter, executive vice president and general manager of AI and customer experience (CX) solutions at RingCentral. In practice, this means that critical nuance is often filtered, summarized, or lost entirely by the time it reaches the system.

Hostetter’s criticism is not that CRMs are outdated. It’s that they linger, often reflecting what a person decided to type after the conversation, not what they actually revealed.

A CRM may show an account as stable, while a sales call reveals uncertainty about budget approval. It can reflect confidence, while a conversation with the customer signals fatigue or second thoughts. These gaps become more important as sales cycles lengthen and buying committees grow.

Industry analysts watch the industry shift in similar terms, even if they describe it differently. In one Forrester blog post on the evolution of sales technology categories, the company describes how modern revenue platforms are increasingly trying to “unify conversational intelligence, pipeline analytics and forecasting” into a single workflow experience, precisely because structured systems alone are lagging behind how decisions unfold.

This is the real pressure point here. Businesses aren’t chasing conversational AI because they need better documentation of sales calls. They do this because they need warnings about deals at risk before those deals show up as problems in CRM.

Early Signs, Mixed Proof

There is no shortage of excitement around conversational artificial intelligence. Businesses are developing tools that analyze voice, chat and video interactions for surface patterns that people might miss, from changes in sentiment to shifts in engagement.

RingCentral’s internal research, conducted with Opinium and scheduled for public release this year, shows that organizations using AI agents are already seeing gains in productivity, customer experience and workflow speed. These results echo what many businesses report anecdotally: Conversational analytics can improve guidance, reduce manual work, and create more consistent follow-up.

What remains harder to prove, however, is whether this information actually translates into better revenue forecasts. Many marketers imply that better conversational intelligence leads to better predictions. But fewer provide public, independently verifiable evidence that conversational AI has improved forecast accuracy in a way that financial leaders would bet on without hesitation.

This challenge is not unique to RingCentral or chat AI. It reflects a broader AI business reality. Gartner has repeatedly warned that while AI adoption is widespread, many projects are struggling to move from pilot value to system-level impact. In a recent outlook, Gartner predicted this over 40% of AI projects it could be abandoned by 2027 due to “escalating costs, unclear business value or insufficient risk controls.”

In other words, while it’s easy for businesses to find patterns in conversations, trusting those patterns when there’s real money at stake remains a challenge.

Augment, Not Replace, CRM

Despite the hype surrounding AI agents, businesses aren’t scrambling to abandon their CRMs. Instead, they’re testing how conversational intelligence can power existing systems without breaking them.

Hostetter is clear on this point: Conversational AI is not meant to displace systems of record. It is intended to inform them. Structured data it still anchors forecasting and reporting, but conversational data adds a level of immediacy that CRMs were never designed to capture.

RingCentral’s research builds heavily on this point, arguing that fragmentation is the limitation holding back AI’s broader impact, not a lack of interest. The structure of his research also matches where business adoption pressure is increasing. Leaders want reliability more than innovation — systems that behave predictably and integrate cleanly, not tools that just sound human.

This preference aligns with Gartner’s warning about cost, value and risk controls, and also aligns with how the conversational AI market is expanding. IDC, for example, forecasts AI chat software service development to revenues of over $31.9 billion by 2028, with a reported CAGR of over 40%, highlighting that businesses are spending here because they expect to become fundamental.

The Takeaway

However, the point is not that voice AI is magical. It’s that conversation hides the “why” behind a decision, and businesses are tired of systems that only store the “what.” If conversational AI becomes valuable to forecasting, it will be because it helps teams detect change early, plug that signal into the pipeline, and drive it into decision-making without turning every encounter into an argument about whether or not the AI ​​can be trusted.

Businesses testing conversational AI aren’t looking for certainty. They’re looking for honesty — a system that can tell them when their assumptions about a deal no longer match reality. If that’s all the conversational AI offers, it might be enough. And that’s because the alternative isn’t a better prediction. It’s another quarterly miss that no one saw coming, even though the signs were there all along.