Earlier this year, Salesforce announced the planned retirement of Open CTI. Now, the integration is in maintenance mode and is scheduled for official retirement on February 28, 2028. Salesforce is now encouraging customers to migrate their telephony operations to Salesforce Voice as soon as feasibly possible.
On the surface, it might seem like a straightforward case of organizations having to move on from an increasingly legacy system. But behind that, there’s a more fundamental transformation at play. That’s because the transition enables organizations to put AI to work in the contact center in a more far-reaching and fundamental way.
From Open CTI to Salesforce Voice… and AI
Open CTI is an integration that enables organizations to surface information about customer interactions in Salesforce. It has been a linchpin of the Salesforce ecosystem for almost a decade and a half, being first introduced in 2012. Since then, organizations have relied on it to bring softphone features like call controls, logs, and click-to-dial functionality into the Salesforce interface from external systems.
Effectively, Open CTI acts as a bridge between Salesforce and external telephony systems, unified communications platforms, and contact center solutions. The integration has worked well enough for several years – but realistically, it was built for a different era. Now, customers want to move away from the ‘bolt-on’ model and limited integration of the interaction records. Today, customer experience centers on immediacy, personalization, and preference. This creates a need for a single, integrated base of operations for customer interactions.
There are several crucial benefits here. The first is the most obvious – it avoids the need to manage and maintain myriad separate datasets across different systems. But perhaps more importantly, it also enables customer data to fuel real-time actions and insights, rather than waiting for it to be updated after a call is complete.
This isn’t just a nice-to-have. In fact, the real-time exchange of information is fundamental for organizations looking to integrate agentic and generative AI functionality (such as Agentforce) into their contact centers. This is almost certainly one of the reasons that Salesforce is moving its customers toward Salesforce Voice.
AI in the Contact Center: How to Save Costs and Improve Outcomes
To understand the impact of these changes, Vonage partnered with SF Ben and leading researcher Metia to understand the state of migration, including drivers, timelines, and benefits – both expected and realized. This involved a survey of 360 Salesforce Open CTI and Salesforce Voice customers.
According to our research, 55% of respondents report that their migration from Open CTI is either being actively explored or underway. With two years to go, there’s still time to think carefully about what this transition means for your organization. But it’s important to get started sooner rather than later.
Understandably, the shift to Salesforce Voice may appear daunting for organizations that have relied on it for years. But respondents who made the switch to Salesforce Voice show that this is an opportunity as well as an imperative. In doing so, you can unlock a whole range of AI-powered contact center functionality and benefits. Done right, that has the potential to speed up resolution, reduce after-call work, and deliver a faster and more consistent customer experience.
But what does an AI-powered contact center look like? And how can it help save costs and improve outcomes for Salesforce-based customer service teams? Here, we discuss six key use cases identified through our research:
1. Real-Time Compliance
Whether the customer is speaking to a human expert or an AI, it’s important that the agent sticks to the script, in order to keep interactions accurate and consistent between different customers. This helps to reduce the danger of compliance breaches and customers being left with inaccurate information.
AI can help with this by cross-referencing the agent’s conversation against company guidelines and compliance documentation. Then, it can offer real-time prompts and guidance to help the agent deliver the right information to the customer in the right way – as they communicate. Salesforce Voice respondents reported 18% higher use of “call quality assurance” than their Open CTI peers.
2. Automated Knowledge Retrieval
Ideally, you want to avoid agents having to search for information mid-call, wherever possible. In the end, it just extends call times, causes friction in the customer experience, and increases the risk of incorrect information.
Historically, this has been a difficult problem to manage. But with AI, we can enable automated knowledge retrieval – which can significantly reduce this issue. It does this by analyzing the conversation and context in real-time and automatically surfacing the answer to the customer’s question. It can also offer help articles, policy documents, or troubleshooting steps.
This accelerates time-to-resolution and helps experts to contain the customer interaction within the call or message thread, by showing agents the information they need, when they need it. No more searching for information mid-call or having to follow up with customers when the interaction is complete. Here, we saw an unsurprising 76% of respondents who were using agentic AI either using or implementing “automated knowledge retrieval”.
3. Handling Routine Tasks
Many routine requests don’t have to go through a customer service agent. With the right self-service functionality, queries like address changes, refunds, or appointment bookings can be managed asynchronously.
Straightforward queries like this are perfect use cases for agentic AI. In many cases, the task can be executed entirely autonomously by the AI agent. This is particularly the case if the request comes via an online form or through a chatbot. And if the customer is already on the phone with a human expert, the AI agent can complete the request in the background while the conversation continues.
This ultimately ensures that human experts have the time and space to deal with complex and high-value interactions – where their expertise is best-placed. It also helps to reduce your cost-per-interaction and expand your contact center capacity, without scaling costs. Within our survey results, Salesforce Voice customers again showed greater adoption over those constrained by Open CTI, with 28% higher adoption of “end-to-end handling of transactional workflows”.
4. Automating Post-Call Tasks
Many contact centers keep a firm eye on metrics like call duration and time-to-resolution. When it comes to customer service, time really is money. But these often fail to capture those post-call tasks that also take up an agent’s time. These are some of the biggest hidden costs in the contact center.
Again, AI has a huge role to play here, by taking on straightforward tasks like compiling call notes, populating CRM fields, and assigning follow-up actions. This also reduces the risk of human error, such as agents promising to get back to the customer and then forgetting.
Here several results demonstrate the importance and benefits associated with advancing this use case as a priority. In terms of adoption, Salesforce Voice respondents were 81% more likely to have implemented “automated post-call work”. At the same time, they were 31% more likely to rank “real-time ‘co-pilot’ guidance for human agents” highest as an actual or expected outcome, in the context of their Agentforce Service / Service Cloud workflows.
5. Proactive Customer Service
Traditionally, contact centers are reactive: Customers report an issue and agents resolve it. But surely, the better approach is to avoid the problem from arising in the first place.
There are many ways to do this. By monitoring metrics like failed transactions, delayed deliveries, or usage anomalies, it’s often possible to identify issues before the customer does. But until now, customer service teams could rarely justify diverting resources from the front-line in order to monitor these issues and manage the consequences.
AI is particularly good at recognizing and flagging these problems, particularly any models with anomaly analysis capabilities. Once issues have been identified, there are essentially three options available:
- Resolve the issue: Just like in section three (above), AI agents can resolve straightforward issues without requiring a human agent, then notify the customer before they realize.
- Assign tasks to a human: For more complex issues, the agent can delegate follow-up actions to a human. This means the resolution can be managed without the need for customers to get in touch – reducing call times and improving the customer experience.
- Follow up with the customer: When more input is required from the customer, the agent can get in touch proactively to notify them of the issue and suggest follow-up actions. Examples here could include failed payments or deliveries.
As well as saving time, this also improves the customer experience. When issues occur, it’s much better to be ahead of the problem, rather than waiting for the customer to notice and complain. Again, unsurprisingly, Salesforce Voice customers maintain an advantage over Open CTI peers, with 28% higher adoption of “proactive customer service”.
6. Agentic Fraud Prevention
Fraud is one of the biggest issues that contact centers have to solve. If your customer isn’t who they say they are, you risk lost funds, compliance breaches, and angry customers.
Many contact centers rely on methods like security questions to manage this risk. Even when this works, it creates a huge bottleneck and friction at the start of every call, as customers can spend significant time trying to recall or find the right answer. This has a direct impact on call durations and time-to-resolution.
Instead, AI can monitor live call data alongside telephony and communication network signals to identify potential fraudulent behavior. This could include details like where the call is coming from, whether the SIM was recently changed, and whether the caller has a standard phone number or an IP phone. The best way to do this is to take information directly from telecoms carriers – which you can do with products like Vonage Agentforce Identity Insights and Fraud Detection.
Otherwise, AI agents can analyze information that you already have access to, such as whether the customer is contacting you with the same number or email address that you have registered for them. This shouldn’t replace customer verification – but it can help flag the biggest fraud risks and reduce the number of security questions required for the lowest-risk interactions. As with Automated Knowledge Retrieval, agentic enterprises such as those that combine Salesforce Voice and Agentforce are more likely to adopt Agentic Fraud Detection – in fact, 2.6x more likely.
Salesforce Voice: An Opportunity and a Requirement
Trepidation about the migration from Open CTI is understandable – but few would disagree that the contact center and customer experience will be measurably better on the other side of this journey.
Indeed, organizations that have already deployed agentic AI in the contact center report a 30% reduction in average handle time and 20% increase in first-contact resolution in the first 12 months.
These numbers come from an exclusive research report compiled by the Vonage team over the last few months, looking specifically at the Open CTI to Salesforce Voice transition. We surveyed business leaders from across the industry to compile new data on readiness, migration pressure, and decision risk for Salesforce organizations.
If you want to understand more about the migration process and what challenges to overcome, this report has the answers you need.
Check out the full research today.
