Salesforce is retiring Open CTI. If your organization leverages it as part of your voice/telephony strategy, the planning window is open….and shorter than it looks.
In February 2026, Salesforce announced that Open CTI, the popular, but very dated JavaScript-based telephony integration framework that has quietly powered Salesforce voice integrations since 2012 will be end-of-lifed February 28, 2028. After that date, Open CTI implementations will no longer function. Not “may experience degraded performance.” Not “will enter an unsupported state.” Will not function.If your organization has dozens, hundreds….thousands of Salesforce users leveraging CTI for your voice to Salesforce integration (for example, your outbound Sales team or an inbound Contact Center), this announcement probably affects you directly and there is a good chance that it has not yet landed with the urgency it deserves.
That is the gap this post is written to help close.
What Salesforce Open CTI Actually Is — and Why It Matters in the Enterprise
Open CTI is the mechanism through which your telephony platform lives inside Salesforce. Every screen pop that fires when a call arrives, every click-to-dial your agents use, every call log that writes automatically to the contact record – it is probably Open CTI.
If you’re running Genesys, Five9, Amazon Connect, NICE, Talkdesk, Vonage, Cisco, or most other CCaaS platforms in a Salesforce environment and do not have SCV – Open CTI is probably the invisible infrastructure underneath it.
Open CTI is also now in maintenance only mode. Salesforce is not building new features for it. New Agentforce Service organizations cannot provision it at all. The framework is aging out, and the replacement, Service Cloud Voice, is architecturally different in ways that matter.
The scale of affected deployments is very large because Open CTI became a standard well over a decade ago. It worked. Organizations built on top of it. Custom softphone layouts, IVR screen pop logic, post-call automation triggers, reporting pipelines; fourteen years of integrations compounded on a framework that is now in its final chapter.
Why 24 Months Is Not as Generous as It Sounds
The instinct is to treat a 2028 deadline as a 2027 problem. That instinct is the risk.
Enterprise contact center migrations have procurement timelines that compress the available runway significantly. A realistic sequencing looks like this: three to six months for vendor evaluation and architecture decisions, two to four months for design and configuration, six to eighteen months for phased implementation in complex multi-CCaaS or multi-region environments. Add the change management overhead of retraining agents at scale, and you are looking at a minimum of twelve months from decision to production, and that assumes a project that does not hit the standard complications of enterprise deployment.
Organizations that begin planning in 2027 will be running their migrations under deadline pressure, competing for partner resources during what will be peak demand across the ecosystem, and making architectural decisions in compressed timeframes. The organizations that begin planning now have something the 2027 starters will not: the luxury of making this an architecture decision instead of a fire drill.
There is a second risk that is less discussed. As the deadline approaches, Salesforce implementation partners with genuine Service Cloud Voice depth and experience will have full books. The organizations that move early get thoughtful engagement. The ones that move late get the leftover bench.
The Three Paths Forward
Open CTI to Service Cloud Voice is not a single migration path. It is an architectural category with three distinct deployment models, and the choice between them is consequential at enterprise scale.
Service Cloud Voice with Amazon Connect (Native bundle) This is Salesforce’s preferred path and the most deeply integrated option. Amazon Connect serves as the telephony infrastructure; Salesforce manages the provisioning and the agent experience. Real-time transcription, Contact Lens sentiment analysis, Einstein AI features, and Agentforce capabilities all operate with the highest degree of integration in this model. For organizations without an existing CCaaS investment, this is the simplest architectural choice.
The cost model is worth understanding before you sign. Service Cloud Voice licensing sits at the intersection of Salesforce platform licensing and Amazon Connect consumption pricing. Amazon Connect charges per minute used plus per feature enabled. For organizations running hundreds of concurrent agent minutes, the economics of bringing your own AWS account (BYOA) is worth considering. For smaller deployments, the BYOA overhead typically exceeds the savings, so it is the better option in terms of deployment costs and long term cost of ownership (LTCO).
One more thing on Amazon Connect and Agentforce: If you are licensing Agentforce for contact center use alongside Amazon Connect, ensure you understand your total cost structure, SCV licenses, plus Amazon consumption, plus Agentforce usage, plus potentially also Data Cloud usage (voice calls harmonized with web behavior, ecommerce tracking, etc to enable AI Agents).
Service Cloud Voice with Bring Your Own Telephony (BYOT) This model allows organizations to retain their existing CCaaS investment — Genesys Cloud, NICE, Five9, RingCentral, Vonage, Talkdesk, and others, while migrating the Salesforce-facing agent experience to Service Cloud Voice. The telephony infrastructure stays where it is. The Salesforce workspace is upgraded.
For large enterprises with mature CCaaS deployments, active contracts, and complex routing logic, this is often the right intermediate architecture decision. It separates the migration problem from the telephony re-architecture problem. The tradeoff is depth of AI integration: BYOT configurations typically require additional configuration to achieve reporting parity with the Amazon Connect native model, and some features that are automatic in the native model require manual implementation in BYOT. There also may be licensing considerations with the existing vendor to consider.
Critically, BYOT is not a single experience. The quality of the Service Cloud Voice integration varies significantly by telephony provider, and not all providers have invested equally in their SCV connectors. Some, like Genesys Cloud are leading the pack, but that is not universal. Before committing, validate your specific CCaaS vendor’s BYOT implementation against your requirements, particularly around omnichannel routing, real time transcription (to better enable Agentforce) and post-call data capture.
Agentforce Contact Center (Emerging) Launched at Enterprise Connect in March 2026, Agentforce Contact Center is architecturally different from both options above. This is a fully native CCaaS platform built on Salesforce’s Hyperforce infrastructure. The telephony stack, routing, transcription, AI agents, and omnichannel layer all running natively on Salesforce. No integration and one vendor to blame if things don’t work properly.
Start with understanding what you own: The CTI Audit
Before any of the above decisions can be made intelligently, the organization needs an honest inventory of its current CTI dependency. Most organizations have never produced this as a single document. It is almost always more extensive than expected.
The audit has four components:
Integration mapping. Identify every place Open CTI is invoked: softphone layouts, custom CTI adapters, screen pop configurations, click-to-dial implementations, call logging automation, IVR data pass-through. For organizations that have customized heavily, and most large orgs are, this is not a trivial exercise.
Coupling assessment. Determine which integrations are modular and which are tightly coupled to CTI-specific behaviors. Modular integrations translate relatively cleanly. Tightly coupled custom logic needs redesign, not just migration.
Downstream dependency mapping. CTI integrations rarely exist in isolation. Call data feeds reporting pipelines. Screen pop logic triggers workflows. Post-call automation writes to objects that feed dashboards. Identify every downstream system that depends on the data CTI currently produces, and validate that Service Cloud Voice produces equivalent data in equivalent form.
License and contract review. What does your current CCaaS contract say about Salesforce integration? What are the terms on exit? If you’re evaluating a CCaaS switch alongside the CTI migration – what are the implications for existing commitments?
This audit is Phase 1. No architecture decision should precede it. Organizations that skip it tend to discover mid-migration that a custom integration they forgot about is load-bearing.
The Real Risk Is Not Missing the Deadline
The February 2028 date will almost certainly generate significant partner capacity pressure in the second half of 2027, just as it did with Process Builder, Lightning and other major Salesforce deprecations. But the risk for organizations operating at enterprise scale is not missing the deadline. It is the opportunity cost of not treating this migration as an architecture decision.
Open CTI was a connector. Service Cloud Voice is a platform. The organizations that approach this migration as a lift-and-shift, (replicate current telephony behavior in the new framework, check the box, move on), will have met the deadline and missed the point. The ones that use this forcing function to make deliberate decisions about where telephony lives architecturally, what AI (Agentforce) capabilities they are building toward, and how they draw the ownership line between CCaaS infrastructure and CRM intelligence, those organizations will be in a structurally different position in 2028 than their peers who just kept the phone ringing. These decisions are more than how to pop the appropriate screen for an inbound call, they are about how AI Agents are going to reduce costs and improve the customer experience in the long term.
The deadline is real. The planning window is now. The question worth asking before your next vendor conversation is not “how do we migrate from CTI to SCV” but “what contact center architecture are we building toward, and does this migration accelerate that or just perpetuate the current state?”
Ateko is the only partner in this conversation who sits on both sides of the ownership matrix. Our Salesforce practice has been implementing post-CTI architectures since SCV was released. Bell, our parent company, is a national carrier with decades of enterprise contact center infrastructure deployment, the pipes and the platform, all under one roof. If you would like a independent opinion before platform vendors start shaping your architecture, that’s a conversation worth having before the RFP goes out.
