Deutsche Telekom launches Magenta AI Call Assistant for voice

Deutsche Telekom unveils Magenta AI voice assistant

• Deutsche Telekom showcased at MWC an AI assistant that “inserts” itself into the voice network to help during phone calls.
• Magenta AI Call Assistant will offer live translation, conversation summaries, and contextual assistance under the command “Hey Magenta”.
• The rollout will begin in Germany “later this year” and will then expand to all the group’s markets, with support for up to 50 languages within 12 months.

What was announced is based on a public demonstration and executives’ statements during a press conference at MWC.

AI assistant in calls
– What it is: an AI assistant integrated into the voice network (not just in an app) to help “during” the call.
– Where it was shown: Mobile World Congress (MWC), at a press conference with a live demonstration.
– Who it was developed with: Deutsche Telekom indicated that the solution was developed together with ElevenLabs.
– Timeliness signal: the announcement was presented at MWC 2026 and, as of now, the fine commercial details have yet to be published.

Launch in Germany

Deutsche Telekom is preparing to launch Magenta AI Call Assistant in Germany, an artificial intelligence service designed to operate within its voice infrastructure and assist the user in real time during a call. The company demonstrated it at a press conference at the Mobile World Congress (MWC), in an attempt to revalue the traditional call in the midst of the era of messaging and digital assistants.

Abdu Mudesir, a member of the board responsible for product and technology, argued that this bet can “bring voice calls back from the ashes,” and linked it to network modernization: those who have “cloudified” the voice infrastructure, he maintained, are in a better position to integrate innovations directly into the core of the network.

The operator did not specify an exact launch date nor did it yet detail the commercial model, including whether it will carry a premium price.

Rollout in Germany without details
– Stated timeframe: the rollout in Germany was announced for “later this year,” with no confirmed day/month.
– What was not confirmed: the commercial model was not detailed (for example, whether it will have a premium price).
– Information context: the details come from a public demonstration and statements at a press conference at MWC.

Features of the AI voice assistant

Magenta AI Call Assistant was presented as a set of functions that activate during the conversation to provide immediate utility: from breaking language barriers

until turning a call into an organizational tool.

Real-time language translation

The most visible feature is live translation during the call. Deutsche Telekom indicated that the service will be expanded over the next 12 months to cover “up to 50 languages,” pointing to both personal and professional use, especially in travel contexts, customer service, or coordination between international teams.

Conversation summaries

The assistant will also be able to generate conversation summaries, a capability aimed at reducing post-call friction: remembering agreements, tasks, or logistical details without relying on manual notes. In practice, this feature seeks to bring voice calls closer to the productivity that is now associated with chats and collaborative tools.

Contextual assistance

The third pillar is “contextual assistance.” According to the demo, by saying “Hey Magenta” the user can ask for help related to what is being discussed: for example, suggesting a restaurant when dinner plans are being discussed or checking availability in real time.

In the MWC scenario, Mudesir did a demo together with Mati Staniszewski, CEO of ElevenLabs, in which they used the assistant to coordinate a dinner while Staniszewski said he was running late to the demo.

Assistance in Live Calls
1) Start a normal call (without changing apps or devices, as presented).
2) When you need help, invoke the assistant with “Hey Magenta”.
3) Ask for a specific action:
– “Translate this…” (live translation)
– “Summarize what we agreed…” (summary)
– “Recommend a restaurant…” or “Check availability…” (contextual assistance)
4) Verify the result on the spot (for example, that the translation or summary reflects what was said) and continue the conversation.
Practical checkpoint: if it isn’t invoked/activated, the assistant shouldn’t intervene; if it is invoked, it’s advisable to confirm with the other person that both understand the change (especially in translation).

Expansion of the service to other markets

After the debut in Germany, Deutsche Telekom plans to bring Magenta AI Call Assistant to “all markets” where it operates. The communicated roadmap focuses, in the short term, on expanding languages—up to 50 in a year—and on adding “more services” later, without specifying which ones.

For now, the company has not provided an exact launch date or commercial terms (for example, whether it will carry a premium price).

The strategy fits with the company’s positioning around AI applied to consumer products: the previous year, Deutsche Telekom had already presented an “AIphone” in collaboration with Perplexity, and at this MWC it also showed a concept to extend these capabilities to smart glasses.

Compatibility with older devices

One of the most striking messages of the announcement was the promise of universal compatibility. Mudesir assured that the assistant should work “on any phone,” even on classic devices like the Nokia 3310.

If it materializes as described, the approach suggests that the differentiating value does not lie in an app or in a specific handset, but in network-level integration: the assistant as a service layer over the call, accessible without requiring modern hardware.

Compatibility on any phone
– Key claim of the announcement: compatibility “on any phone,” with the Nokia 3310 as an explicit example.
– Operational implication (as presented): the “place” where the assistant lives would be the network/voice service, not the device.
– What is worth confirming when it launches: whether it applies to prepaid/postpaid, roaming, VoLTE/VoWiFi, and whether there are limitations by handset or by call type.

Privacy and activation of content processing

Deutsche Telekom addressed the most sensitive point directly: privacy in AI-assisted calls. According to the company, the system will only begin processing spoken content when the customer activates it; until then, the content of conversations will not be stored or analyzed.

In parallel, Mudesir emphasized that AI will also have a role in mobile network security, in particular to detect

voice deepfakes and other fraud. In that context, he posed an emerging challenge: with AI “agents” increasingly present, how to identify when one is interacting with an agent and not a person.

Key advantages and doubts
Pros
– Explicit control: processing starts only when the customer activates it.
– Less “by default” exposure: if it isn’t activated, the conversation should not be stored or analyzed (as communicated).
Limitations / reasonable questions (not yet detailed publicly)
– What happens when it is activated: whether audio or transcripts are saved, for how long, and for what purpose.
– Where it is processed: whether processing happens on the network, in the cloud, or in a combination, and how that changes by country.
– Signals for the other party: how it is communicated (or not) to the interlocutor that there is assistance/AI on the call.

The Revolution of Customer Experience in Telecommunications

The emergence of AI assistants in the traditional call reopens a strategic question for the sector: how to turn a mature service—voice—into a differentiated, useful, and secure experience, without adding complexity for the user.

Digital Transformation in the Telecommunications Sector

Deutsche Telekom’s proposal is based on an underlying trend: the migration of historically telecommunications functions toward more flexible architectures, with cloud components and rapid iteration capability. In that framework, voice stops being just connectivity and becomes a software-“augmented” channel.

The Importance of Automation in Customer Service

Although the announcement focuses on calls between people, the same capabilities—translation, summarization, context—foreshadow impacts on customer service: better-informed agents, shorter handling time, and continuity across channels. Automation, when applied well, can reduce friction without dehumanizing the service.

Current Challenges and Future Opportunities

The immediate challenge is to balance usefulness and trust: activating features without intruding on the conversation, and offering transparency about when and how AI intervenes. The opportunity, by contrast, is to reposition the call as an intelligent channel: more accessible, more productive, and potentially more secure against fraud.

The Role of Artificial Intelligence in Improving the Customer Experience

AI applied to voice can turn an ephemeral interaction into actionable information: summarized agreements, reduced language barriers, and decisions facilitated in real time.If execution is consistent, the call can regain relevance as an everyday tool, not just as a last resort.

Strategies for Customer Retention in a Competitive Market

In saturated markets, differentiation usually comes from experience, not coverage. A network-integrated voice assistant—compatible with any phone—aims at an advantage that is hard to replicate with handsets or apps alone: a cross-cutting service that accompanies the customer without requiring them to change habits or devices.

Criteria for Smart Calls
Usefulness → Does it solve something “during” the call?
– Live translation, summaries, and contextual help are useful if they reduce friction in the moment (not afterward).
Trust → Does the user feel in control?
– Explicit activation and clarity about what is processed/when are decisive for it to be used.
Adoption → Does it work without changing habits?
– The promise of working on any phone (even older models) lowers barriers and speeds up trial.
Differentiation → Is it hard to copy?
– If the advantage lies in network-level integration and in operations (quality, latency, privacy), it may be more defensible than a simple app.

Innovation in calling involves integrating AI into the network without requiring new devices, but with transparency and control. This type of announcement reinforces an idea we see repeated in telecom operations: the best results usually come from hybrid models, where automation covers the predictable and human oversight retains control when it matters. At Suricata Cx we observe these trends to translate them into omnichannel operational flows with traceability, escalation, and resolution metrics.

This text is based solely on information publicly disclosed by Deutsche Telekom during a demonstration and press conference at MWC 2026. At that time, some details were not communicated, such as the exact date, the price, and operational details by country. Availability and terms may vary as the commercial launch approaches and new updates are published.