Reggie Scales, President and Head of Business Unit Applications at Vonage, is a seasoned technology and communications executive with more than two decades of experience leading sales, operations, go-to-market strategy, and business growth. Since joining Vonage in 2018, he has held senior leadership responsibility within the company’s applications business. Previously, Scales served as Senior Vice President for the Western U.S. at Masergy, Vice President and General Manager of the Central Region at Comcast Business, and Senior Vice President of Sales and Marketing at Wilcon. Earlier in his career, he spent more than 12 years at PAETEC, progressing through several executive positions, including President of Sales and Service, where he directed sales and business operations for a $1 billion region with approximately 500 employees. His career has centered on scaling communications businesses, building high-performing teams, and developing enterprise sales and customer engagement strategies.
Vonage is a global enterprise technology company and part of Ericsson, providing communications APIs, unified communications, and contact center solutions to more than 100,000 businesses worldwide. Its platform enables organizations to embed voice, video, messaging, authentication, and other communications capabilities directly into applications and customer experiences, while its business communications and contact center products support employee collaboration and customer engagement. Vonage is increasingly incorporating artificial intelligence and mobile network intelligence into its platform, giving businesses and developers tools to create more personalized, secure, and automated communications experiences across digital channels
Your career has spanned sales, service, regional operations, communications infrastructure, and now leadership of Vonage’s applications business. How has that operating experience shaped your approach to AI adoption, particularly the need to balance innovation with measurable customer and business outcomes?
I have spent more than 25 years in communications, and each stage of my career has reinforced the importance of connecting technology investments to a clear business need. I started by selling long-distance services, then moved through local communications, data networks, cloud computing, and now AI. I have had the opportunity to see several major shifts in our industry, and one lesson has stayed consistent: new technology delivers lasting value when it solves a real problem for customers and the business.
I apply that same thinking to AI. In a contact center, the starting point should be the outcome you are trying to improve. That might be helping customers resolve routine questions faster, giving employees better information during a conversation, reducing unnecessary steps, or making it easier to move between channels without losing context.
My current role spans product, engineering, sales, partnerships and customer support, which also shapes how I think about adoption. A capability can be impressive technically, but enterprises still have to consider how it fits into existing operations, how easily employees and customers can use it and how they will measure whether it is working. That connection between technology, execution and outcomes is especially important with AI because the pace of change is so fast.
Many enterprises are eager to introduce generative and agentic AI into their contact centers. What foundational work must they complete across data, workflow design, systems integration, and governance before attempting to scale beyond an initial pilot?
Enterprises need a strong foundation before they expand AI across the contact center. A pilot can work in a controlled environment with a narrow set of data and a limited workflow. Scaling it means connecting the AI to more systems, more customer information, and more parts of the business, which introduces additional complexity.
Data is one of the first areas to address. Organizations need to understand where their customer information lives, whether it is accurate and current, and who has permission to use it. Workflow design matters just as much. Companies should understand the process the AI is entering, what decisions it can make and where a person needs to become involved.
Integration is another major piece. Customers communicate through voice, messaging, video and other channels, while employees may rely on several systems behind the scenes. If those environments remain disconnected, adding AI can create another layer of fragmentation.
Governance should be established from the beginning as well. Enterprises need clear rules for privacy, security, access, and human oversight. They should also define how they will evaluate the AI over time. Moving beyond a pilot requires confidence that the technology can perform consistently as the number of interactions and the range of situations increases.
Customer information is often fragmented across customer relationship management systems, call transcripts, messaging histories, and other enterprise platforms. What does a genuinely unified customer context layer require, and how can organizations prevent AI systems from acting on incomplete or outdated information?
A useful customer context layer starts with connecting the information that explains who the customer is, what has already happened in their journey, and what they are trying to accomplish now. That can include CRM information, call transcripts, messaging histories, account details, previous service interactions, and other relevant records.
Many companies still manage those pieces across separate systems, and that creates problems, with or without AI. A customer might begin a conversation in one channel, move to another, and then have to repeat information because the second system does not have the history of the first interaction. AI makes solving that fragmentation more important because an agent can only make a useful recommendation or take appropriate action when it has reliable context.
Organizations need to know which systems are authoritative for different types of information and have processes for keeping that information current. They also need safeguards for situations where the AI does not have enough context to proceed confidently.
If information is missing, conflicting or outdated, the AI should be able to recognize that limitation and ask for clarification or involve a human. That is particularly important as AI moves from answering questions toward taking actions for customers.
What should an effective handoff from a virtual agent to a human advisor look like? Which information about the customer’s intent, previous conversation, completed actions, and unresolved issue should transfer so the customer does not have to start again?
The customer should experience a seamless handoff as a continuation of the same conversation. By the time a human advisor joins the interaction, they should understand why the customer reached out, what has already been discussed, what the virtual agent has completed, and what still needs attention.
At a minimum, the handoff should include the customer’s intent, the relevant conversation history, account information needed for the interaction, any steps or verification that have already been completed, and the reason the virtual agent escalated the issue. If the customer has already explained the problem or answered a question, the advisor should have that information readily available.
That continuity matters because one of the biggest sources of frustration in customer service is having to start over. AI should help reduce that friction rather than add more steps.
As virtual agents become capable of handling a larger share of routine interactions, the quality of the transition to a person becomes even more important. There will continue to be situations that require empathy, judgment, creativity, or more nuanced problem-solving. Companies need to design for that from the beginning and give employees enough context to pick up the conversation effectively from an AI agent when those situations arise.
The next phase of customer experience AI is expected to focus less on standalone tools and more on intelligence embedded directly into existing workflows. How does this change the role AI plays in the contact center, and how should enterprises decide which tasks to automate completely versus which should remain human-led?
Embedding AI into existing workflows is highly effective because employees and customers do not have to move into a separate experience to reap its benefits. AI can support what is already happening in the contact center by helping an employee find information during a conversation, summarizing an interaction, suggesting a next step or completing a routine process for a customer.
When deciding what to automate, enterprises should look closely at the nature and consequences of the task. Repetitive interactions with clear rules and predictable outcomes are strong candidates for greater automation. A straightforward account inquiry, for example, is very different from a conversation involving a sensitive complaint, an unusual circumstance or a decision that could have a significant impact on the customer.
Employees will continue to play an important role where empathy, judgment, and creative problem-solving are required. AI can also support those employees rather than simply taking over an interaction.
I would evaluate automation based on the experience it produces. If a process becomes faster but leads to more repeat contacts, more escalations, or greater customer frustration, the company needs to revisit how that workflow was designed.
Vonage recently introduced industry-specific AI agents for healthcare, financial services, and retail contact centers. Why is vertical specialization becoming important, and how should an enterprise determine whether an AI agent genuinely understands its industry’s workflows, terminology, and regulatory requirements?
The way a customer engages with a healthcare provider can be very different from the way someone interacts with a financial institution or retailer. The underlying contact center technology may have similarities, but the language, workflows, customer expectations, and requirements surrounding those interactions can vary significantly.
That makes industry context increasingly important as AI agents become capable of doing more than answering basic questions. An agent needs to understand the types of requests customers commonly make, the terminology employees and customers use, the systems involved in completing those requests, and the boundaries around what it should be allowed to do.
Enterprises should evaluate industry-specific AI using scenarios taken from their actual operations. That means testing common requests along with exceptions and more complicated situations. They should look at whether the agent follows the expected workflow, understands the language customers use, and knows when it should involve a person.
Organizations also have to assess how privacy, security and regulatory requirements are handled. Those considerations should be part of the design and evaluation process rather than something addressed after an agent is already interacting with customers.
Ultimately, industry knowledge becomes valuable when it translates into more accurate, appropriate, and useful customer interactions.
As AI agents progress from answering questions to completing transactions and taking actions on behalf of customers, what safeguards are required around permissions, privacy, hallucinations, auditability, and human oversight?
The level of oversight should increase as AI agents are given more responsibility. There is a significant difference between an agent providing general information and an agent making an account change, completing a transaction or taking another action that directly affects a customer.
Permissions are a key part of that. An AI agent should only have access to the information and systems required for the task it has been authorized to perform. Identity verification, access controls and clear rules for how customer data is used are also important. Customers should have transparency into how their information is being handled.
Organizations also need a record of what happened during an interaction. As agents begin taking actions, enterprises should be able to understand what information was used, what action was completed and where an issue occurred if something goes wrong. That becomes particularly important in sensitive or regulated environments.
Human oversight should reflect the risk of the task. Some routine actions may be appropriate for an AI agent to complete independently. More consequential situations may require review or approval.
The AI also needs to recognize uncertainty. When information is incomplete, conflicting or outside the boundaries it has been given, the safest course is to pause, ask for additional information or involve a person rather than proceed based on an unsupported assumption.
Contact center AI projects are often evaluated through cost savings, call deflection, or reductions in average handling time. Which additional metrics should leaders use to determine whether AI is improving first-contact resolution, customer satisfaction, employee experience, revenue, and long-term trust?
Efficiency metrics matter, but leaders should look beyond how quickly or cheaply an interaction was handled. A short interaction is not particularly valuable if the customer has to contact the company again because the issue was never fully resolved.
First-contact resolution is an important measure for that reason. Repeat contact rates and escalation rates can also tell leaders whether AI is genuinely helping customers or simply moving the problem elsewhere. Customer satisfaction should remain central because the experience ultimately determines whether people continue to trust and choose to do business with a company.
That point is especially important given what we have seen in our customer engagement research. Customers have limited tolerance for poor experiences, so companies should be careful about optimizing one efficiency metric at the expense of the overall relationship.
Employee experience deserves attention as well. If AI is providing better information, handling repetitive work and helping employees solve more complex issues, leaders should be able to see that in employee feedback and performance.
I would also look at broader business outcomes such as customer retention, account growth, and revenue generated or protected through service interactions. Taken together, those measures give leaders a much clearer picture of whether AI is creating value for customers, employees, and the business over time.
Vonage operates across contact centers, unified communications, communications application programming interfaces, and network-powered capabilities through Ericsson. How can these layers work together to help AI agents communicate and act across channels without creating another fragmented technology stack?
Customers do not think about whether an interaction is happening in a contact center platform, or powered by a communications API or the mobile network. They expect to be able to communicate through the channel that works for them and continue the interaction without unnecessary friction. The technology behind that experience has to support that continuity.
Vonage operates across Network APIs, CPaaS, CCaaS and UCaaS, which gives us an opportunity to think about these capabilities as connected parts of the customer experience. As AI agents become more capable, that connection becomes increasingly important because an agent may need to communicate through one channel, access context from another system, and take action somewhere else.
A more integrated approach can also help maintain context as customers move between voice, messaging, and other channels. That same principle applies when an AI interaction moves to an employee to manage.
Our relationship with Ericsson also creates an opportunity to bring communications applications and network capabilities closer together. The goal is to ensure those capabilities are useful within the customer journey without forcing businesses to manage another set of disconnected tools.
Can you please share one example of how Vonage’s contact center, unified communications, API and Ericsson network capabilities can work together in a customer interaction today?
A good example is what happens when a customer calls a financial services contact center to report a suspicious transaction or request an account change.
Before the agent even picks up, Vonage’s Network APIs – powered by Ericsson’s carrier infrastructure – are already at work. The moment the call comes in, our Identity Insights capability checks the caller’s phone number against live mobile network data, including whether the SIM card associated with that number has been recently swapped. That is a strong signal of potential fraud, and it happens invisibly, without any friction for the customer.
That network intelligence flows directly into the contact center through Vonage’s Communications APIs, surfacing as a real-time trust signal inside the agent’s workspace. If the agent is working within Salesforce, that insight appears automatically within Agentforce, so the agent can see whether the caller’s identity has been verified or whether there is a reason to apply additional scrutiny – all before the conversation begins.
At the same time, Vonage Contact Center (VCC) seamlessly integrates with its unified communications solution, Vonage Business Communications (VBC) to connect with the broader organization. If the agent needs to bring in a fraud specialist or a back-office colleague, they can see who is available across the business in real time and transfer the call with full context intact. The customer does not have to repeat themselves, and the specialist joins the conversation already knowing what has happened.
What makes this example meaningful is that none of these layers are operating in isolation. The network is informing the contact center. The APIs are connecting the intelligence to the workflow. The unified communications layer is keeping the right people in the loop. The customer experiences it as a single, continuous interaction — which is exactly what it should feel like.
That is the kind of integration we are working toward across all of our capabilities, and it is already available to customers today.
As AI assumes more routine customer interactions, how do you expect the responsibilities of human contact center professionals to change over the next three to five years, and what should enterprises begin doing now to prepare their people, processes, and leadership?
As AI handles a larger share of routine interactions, I expect contact center professionals to spend more of their time on the situations where human skills matter most. That includes complex problems, unusual circumstances, and conversations that require empathy, judgment or creativity.
Contact center professionals’ roles will also become more closely connected to AI. Employees may use AI to find information, understand the history of an interaction, summarize conversations or identify possible next steps. That means people need to understand how to work with these systems, where to trust them, and when to challenge or override a recommendation.
What’s important is that enterprises begin preparing their teams now and leadership has a role critical role to play. Employees need clarity about why AI is being introduced and how their jobs are expected to evolve. Training should help employees understand how AI fits into their day-to-day responsibilities and what remains their responsibility when AI is involved.
Companies should also involve contact center professionals in implementation because they see customer questions, process gaps, and unusual cases every day. Their experience can help identify where automation will be useful and anticipate where it could create issues.
Over the next three to five years, I expect the strongest contact centers will be those that use AI to make their people more effective while preserving a clear path to human support when customers need it.
Thank you for the great interview, readers who wish to learn more should visit Vonage.
