MODaaS, token economics & agentic AI - Building scalable AI-native operations

Sessions in rooms

Presentation: Running AI like a system: From models to managed services 
This session explores how operators are managing AI as a production system, using modular architectures, and governed deployment models like MODaaS. 

• How telcos decide what runs as a service vs embedded AI 
• How to manage multi-model environments across vendors  
• What breaks first when AI hits production?  
• What a real MODaaS operating model looks like  

  Make AI Stick: The 4 Moves to Run an AI‑First Telco 

• How to move AI out of pilots and into live operations across your business 
• How to redesign workflows and systems so AI drives operational decisions — not just reporting 
• How to make your organisation truly data‑driven, with the right foundations for AI at scale 
• Understand what the most successful adopters are doing differently — and how to apply it 

Christian Livia, CTIO, Integratel Peru (Movistar)

Taming AI spend: Token economics learnings from the frontline 

• From tokens to ROI: How telecoms operators are turning token consumption into measurable business value 
• Cutting cost without cutting capability: Practical lessons from model selection, routing and prompt optimisation to reduce inference spend at scale 
• Building AI cost discipline: What finance, technology and network teams need to know to forecast, govern and optimise rapidly growing AI workloads. 

Mark Austin, Vice President, Data Science, AT&T 
Andy Markus, Chief Data and AI Officer, AT&T