Radisys V.AI launched on 10 September 2026 as an ecosystem of voice and speech AI tools that telecom operators can deploy through Radisys’s existing Engage Digital Platform. The release creates a path for network-delivered assistants, call screening, translation and recording, but it does not mean those services are already available from every carrier.
Radisys V.AI moves AI into the call path
The official Radisys release says V.AI is integrated with Engage Digital Platform, or EDP, which the company introduced in 2022. EDP already handles real-time media in operator environments; the new ecosystem adds partner speech technologies and a development environment around that delivery layer.
The five charter members are ElevenLabs, Hiya, ReadSpeaker, Sensory and Speechmatics. Radisys says their capabilities can support pre-call, in-call and post-call experiences, including fraud protection, call screening, live translation, voice assistants and intelligent recording. Those are possible applications rather than a promise that each operator will offer every feature.
Independent coverage from Open-RAN.org confirmed the ecosystem structure and argued that its commercial test is whether operators can turn network infrastructure into paid services. BTW Media separately highlighted the same boundary: placing AI in a familiar call flow may reduce the need for another app, but it does not prove user demand, reliability or acceptable cost.
The architecture matters because a network-level service can reach ordinary mobile and landline handsets. A user may not need a new application if the operator inserts translation, screening or assistance into the call path. That broader reach also raises the stakes for consent, error handling and data retention.
Deployment choice is useful, but not proof
Radisys says operators can run partner solutions in their own managed infrastructure or use public-cloud services, including speech tools from Microsoft, Google and OpenAI. On-premises deployment may help with data-residency and latency requirements, while cloud services may simplify access to new models. The trade-off depends on the workload, traffic volume and national rules.
Claims about lower latency, stronger privacy or efficient deployment should therefore be tested per configuration. A live-translation service needs end-to-end delay measurements, language accuracy in noisy calls and a clear fallback when confidence falls. A fraud-screening service needs false-positive and false-negative rates, plus a way for legitimate callers to recover.
Operators also need a consent model. Subscribers should know when audio is transcribed, which provider processes it, how long records are retained and whether a human can review the result. Network delivery should not turn a feature into an invisible default simply because it can run without an app.
In plain terms, Radisys V.AI is a distribution and integration layer for telecom voice AI, not a single consumer product. Its value will be proven only when named operators launch specific services with measurable reliability, privacy controls and customer adoption.
| Announcement date | 10 September 2026 |
|---|---|
| Foundation | Radisys Engage Digital Platform |
| Initial partners | ElevenLabs, Hiya, ReadSpeaker, Sensory and Speechmatics |
| Build path | Packaged apps, custom services or V.AI Studio |
| Not disclosed | Operator customers, pricing and market launch dates |
What buyers should test first
A sensible pilot begins with one bounded call flow and a baseline. The operator should measure completion rate, latency, recognition errors, customer opt-outs, escalations and total cost per successful interaction. Results should be broken down by language, device type and network condition rather than compressed into one average.
The controls should be tested as actively as the feature. Teams can verify that transcripts stay within the chosen region, access permissions are enforced, recordings expire on schedule and incidents can be traced across Radisys and the partner service. Contracts should identify who is responsible when the network, model and application layers disagree.
This evidence standard resembles the control questions around Zayo’s agentic networking tools, where permissions matter as much as automation. It also applies to Mahindra Finance’s multilingual voice AI, where real language coverage and escalation determine whether a voice system helps customers.
Everyone else is reporting an ecosystem launch; we are explaining the delivery mechanism and the evidence operators need before turning it into a subscriber service. The distinction prevents a partner list from being mistaken for deployed adoption.
Frequently asked questions
What is Radisys V.AI?
Radisys V.AI is an ecosystem that connects voice and speech AI partner technologies to the Engage Digital Platform for operator-built communication services.
Can subscribers use Radisys V.AI now?
Not as one universal product. Availability depends on whether a telecom operator selects, configures and launches a particular service.
Which companies joined the launch?
Radisys named ElevenLabs, Hiya, ReadSpeaker, Sensory and Speechmatics as charter ecosystem members.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
