Nuance Mix end of support on Genesys Cloud: what ends and your options

Genesys published the deprecation on 4 November 2024. Its notice says that Nuance has announced that Nuance Recognizer as a Service, Nuance Text to Speech as a Service and Nuance Mix Dialog will reach end of life by 31 May 2027, and that on that date Genesys will discontinue support for the Nuance Recognizer as a Service, Nuance TTS and Nuance Mix Bot integrations. Until then, Genesys says customers can continue to use these features.

The notice also set out how service extensions worked: the listed Nuance products could be used without an extension until the end of 2024, and customers could apply to Nuance to extend their service up to 31 May 2027. If your organisation still uses any of these services today, it is running on that extension, so the date is firm. We could not find a separate end-of-life notice on Nuance's or Microsoft's own pages, so this guide relies on Genesys's notice for the date.

Three things are affected, and each shows up differently in Architect. Nuance Mix bots are called from flows with the Call Nuance Mix Bot action; Genesys's documentation describes the integration as calling Nuance Mix bots in Architect message flows, and notes that the feature is PCI DSS compliant, which matters if payments run through it. Nuance Recognizer as a Service provides grammar-based speech recognition in Architect bot flows. Nuance TTS appears as the text-to-speech engine, or as a voice, in Architect flows.

Because prompts and recognition are often shared through reusable tasks and common modules, the safest inventory is a search of every published flow version, not only the main entry points. The list below is the order we work through with clients.

1. Export every Architect flow and common module, including inbound call, in-queue, message, bot and workflow flows.
2. Search for the Call Nuance Mix Bot action and record what each bot does, its intents and its hand-off points.
3. Search for grammar-based recognition and note each grammar and the data it collects.
4. List every prompt that uses a Nuance TTS engine or voice, including dynamic text-to-speech.
5. Map each use to the journeys, queues and reports that depend on it, and to any PCI scope.
6. Record current call and message volumes for each, so the replacement can be sized and costed.

Genesys's notice does not name a replacement; it asks affected customers to work with their sales representative. What Genesys does document is a set of supported bot and speech routes in Genesys Cloud, compared side by side in its channel, bot and touchpoint capabilities article. The right choice depends on what the Nuance service was doing: a simple menu or data capture, a conversational bot, or speech recognition and voices for prompts.

For prompts and voices, Genesys offers its own text-to-speech, including enhanced voices, and supports third-party engines under its bring-your-own-technology model. For bots, the table summarises the main routes. Our guide to native or third-party voice AI on Genesys Cloud compares them in more depth, including where each is limited.

A Nuance replacement is a redesign, not a lift and shift. Intents, grammars and prompts were tuned for the Nuance engine, and a different engine will recognise and speak differently. We plan backwards from 31 May 2027: inventory and design first, then build, then a test period long enough to tune recognition on real traffic, with time left to run old and new side by side and fall back if needed.

Testing deserves the most attention. Recognition accuracy, barge-in, silence handling and hand-off to agents all change with the engine, and the only reliable measure is real calls. We test the new flows against recorded utterances and scripted calls first, then on a share of live traffic, and track containment, transfers and abandoned calls against the Nuance baseline. Where a flow takes payments, the PCI design and assessment need to be reviewed before the change goes live, not after.

Each route is charged differently, so compare them on your own volumes. Genesys bills third-party bots under its bring-your-own-technology rate classes: Nuance Mix voice sits in the same rate class as Amazon Lex and Dialogflow voice, while Nuance Mix messaging sits in another, charged per turn for digital and per minute for voice in 15-second increments. Native Genesys bots and Virtual Agent consume AI Experience tokens instead, and third-party vendors bill their own usage on top.

Our voice AI cost calculator compares these routes at each vendor's published list prices, and our AI tokens guide explains how token use is measured. Both show the date their prices were last checked. The one-off cost of redesign, build and testing usually outweighs a year of usage, so size it in discovery rather than from a rule of thumb.

QVCCS leadership has worked in contact centre technology since 1984, and moving speech applications between engines is familiar ground. We start with a full inventory of Nuance use across your Architect flows, propose a route for each use based on what it does today and what you want it to do next, and run the rebuild, testing and cut-over with a fallback plan, well before the deadline.

Every delivery team member holds, or is completing, Genesys Cloud certification, and we work remotely unless agreed otherwise. We do not sell licences and take no commission, so our recommendation follows your requirements rather than a product. Our pages on third-party bots and on AI Studio and virtual agents describe the platforms in more detail.