How Deutsche Telekom is rewiring telecommunications with AI
Summary
Deutsche Telekom achieved 50% containment rates and NPS improvements by embedding OpenAI's models directly into customer-facing operations, reducing resolution time from hours to minutes while treating AI partnerships as internal team extensions rather than vendor relationships.
Key Takeaways
- Achieve 50% containment rates by deploying AI models at scale across high-volume customer touchpoints (40M annual interactions), targeting specific use cases like live translation, in-call assistance, and post-call summarization.
- Reduce operational execution time from hours to minutes for routine tasks, creating significant OPEX savings and simultaneous customer experience improvements—measure both efficiency and satisfaction metrics.
- Treat AI platform partnerships as embedded team extensions rather than external vendors; prioritize close collaboration and knowledge transfer between your engineers and the AI provider's team for compounding value beyond project scope.
- Design AI implementations with a clear roadmap toward automation parity—focus on use cases where AI can eventually outperform human agents in specific contexts, not across all operations.
- Start with core customer-facing operations where AI can access millions of daily interactions; this volume creates the data and feedback loops necessary to achieve measurable containment and NPS gains at scale.
Related topics
Transcript Excerpt
[music] >> We have 40 million opportunities across Europe every year to either delight or disappoint our customers. So the idea now that we can use opening our models to fundamentally bring that into the core where customers are speaking millions of times every day. Things like live translation, things like in call assistance, things like post call summarization. What we are seeing is containment rates of up to 50% and NPS of about 22 which is not fantastic yet. What would previously have taken previously have taken us hours actually can be done in minutes now and that's a huge saving from us from an up ex perspective but also from a customer experience perspective as well. You provide us the pipe with the intelligence and it's for us to now deploy it inside our organization. I think it's …