The rapid evolution of AI is reshaping the UC sphere like never before.
Already, all the big players have AI-enabled or AI-assisted features in their suite of products that pass on efficiency boosts to their customers.
Yet, the AI race is not over. Just like generative AI was the topic of discussion in 2024, 2025 is coalescing around a new form of AI: Agentic AI.
“This [Agentic AI] is an absolute focus for us,”
John Finch, Global VP Product Marketing at RingCentral told UC Today.
With its ability to operate autonomously, enable end-to-end problem-solving across customer service, internal collaboration, and operational workflows, its scope to service the UC sphere is untold.
Yet with all the buzz around Agentic AI, it can be hard to make sense of where the technology may have the biggest impact, and how to prepare for that.
To find out more about RingCentral’s approach to agentic AI, UC Today sat down with Finch for a one-to-one talk on the technology, its trajectory, the trials in implementation and the company’s ambitions with it.
From Assistants to Autonomous Agents
Like AI in general, Finch sees Agentic AI and its use in the UC sphere as one that is currently evolving.
“I think, in general, where we’re going with this [Agentic AI] is really about how to take it to the next level, right?” Finch says.
“It’s not about the use cases in particular, like, how do I solve my microphones popping in and out? It’s really about how we are taking this to the next level.”
Rather than seeing it as a technology that is due to really take hold of one element of UC, Finch says RingCentral sees AI as an opportunity to transform every organisation into a customer-centric entity.
“From a UCaaS perspective, we could see an AI receptionist or writer,” Finch said. “Yet for a CCaaS solution, those capabilities are equally as helpful; that shows how it’s not just one area that stands to benefit.”
Leveraging it across the whole UC offering, Agentic AI has abilities and features that lend to customer service, internal communications, and cross-departmental interactions.
Yet, Agentic AI augments this by shifting its abilities from a reactive to a proactive system.
Therefore, while virtual assistants and transcription services laid the groundwork for optimisation, Agentic AI applications now focus on “solving problems organisations didn’t know they had,” Finch said.
Thus, the idea behind what Agentic AI can do is understood, but the full extent of what process it can be used for and exactly how it will affect workflows, both customer-facing and internally, are still being discovered.
Yet, Finch believes RingCentral’s approach to Agentic AI will take shape as a deeper presence working alongside employees during their work interactions, in an active listening capacity.
“Having some AI agent that’s working in the background on the app, listening to the conversation and making suggestions,” says Finch.
Indeed, around Q3 of 2024, RingCentral released Telephony solutions that had advanced AI features.
AI Call, for instance, introduced features designed to enhance business phone interactions through intelligent, background-powered tools.
These AI-driven capabilities transform traditional phone calls by providing real-time assistance, transcription, and insights for sales and customer service teams.
The technology includes features like AI Assistant (which takes automatic notes and summarises calls), AI Assist (offering real-time guidance during conversations), predictive dialers, and conversation intelligence tools.
Although not utilising fully autonomous AI systems there, it is an area that RingCentral sees as fertile ground for developing Agentic AI on top of.
“We introduced that, and most of our road map going forward will be focused on AI and generative AI, and those agentic AI capabilities,” Finch said.
Technological Innovation and Governance
Despite the large interest in AI and Agentic AI, one of the things holding back enterprises is implementation, governance and regulation surrounding its use.
For some industries like finance, this is more pronounced. Yet for any organisation operating in the EU, the new EU AI Act has meant that AI use has to meet certain criteria.
When using a UCaaS, customers expect a more turnkey solution, and this equally extends to the AI being used.
Finch says RingCentral is therefore well positioned in this department. The UCaaS uses a combination of proprietary AI models and third-party models for its AI-powered solutions.
“We have our own AI model, a core element of which uses a multi-faceted approach. Generative AI is a part of that,”
Finch said.
Equally, RingCentral keeps close control of what goes in to train the model. Keeping close control of an AI model makes its behaviour more predictable for use.
“It’s contained in a way that we control and that allows for the security and reliability. Built fit for purpose, it’s for business communications; it isn’t an open AI model,” Finch continued.
By not allowing it to feed on the data from the whole internet and instead feeding it business-specific information, Finch asserts RingCentral’s model will not suffer the same severity of ‘hallucinations’ that these other AI models do.
This can give customers the confidence to engage in using Agentic AI without overt fear of falling foul of regulations that relate to things like AI discrimination or bias.
So with these capabilities and safety features present, what does RingCentral see as an area of interest for deploying Agentic AI across its platform?
“If you think about virtual assistants, right or IVAs [Interactive Virtual Agents], we’ve been investing in these for quite some time,” Finch explained.
“The acceleration of these continuing to evolve their capabilities will continue at a very accelerated pace. And we’re going to be introducing some things coming up in a couple of weeks that look at helping clients give their customers a new virtual experience with their brand.”
Although not getting specific, Finch mentioned a “receptionist” capability as something Agentic AI is an “acceleration you will continue to see from us.”
Practical Implementation and Future Outlook
For businesses considering agentic AI adoption, it can all seem very confusing. But as Finch explained, you’re not alone.
“It’s confusing for us as it moves so quickly in the market. So therefore it’s got to be really confusing for companies that are thinking about how to do this.”
He, therefore, recommends a strategic approach to those looking at experimenting with implementation.
“I think, in general, organisations really need to think about where is the impact going to be for their business front and centre. Like, how quickly can you show results and investing in the right areas is really one of those that makes the most sense,” Finch explained.
With the most promising initial areas being customer service and sales, where AI can demonstrably improve key performance indicators like through Net Promoter Scores and customer satisfaction ratings.
Yet beyond this, Finch advocates really getting into the weeds of each department, like legal and finance, as despite it being too complex for many companies currently, it could likely see a major impact.
That is no small task, however. But by doing it piece by piece, then enterprises can begin to think about joining the dots.
“These disconnected systems are the areas that you’re going to see the most, you know, impact right away,” Finch states.
“Then, you can launch into how are we, as an organisation going to invest further into AI, outside of selecting a vendor? Then you can get a UCaaS vendor like RingCentral to go in and bridge these disconnected systems and enhance the real-time interactions in a holistic way.”
As we move further into 2025, agentic AI promises to become an integral component of the UC sphere, offering unprecedented capabilities for businesses willing to embrace this technological evolution.
Integration and governance represent challenges, but by taking an ROI perspective on Agentic AI, and by working with vendors able to alleviate some of the issues, then businesses can feel more confident to approach this next iteration of AI.
