Running AI you can control - Sovereign AI for networks, data and decisions

From 1.1 Million Weekly Conversations to Actionable Insights: How AT&T Built an Enterprise-Scale Transcript Intelligence Platform 

• Enterprise AI at scale requires purpose-built architecture—AT&T replaced vendor platforms with an internal solution processing 1.1M+ weekly interactions with significant cost savings 
• Multi-model AI architecture dramatically reduces token costs while maintaining quality at massive scale 
• Cross-functional collaboration accelerates delivery—6-month timeline vs typical enterprise cycles 

Bo Gu, Director of Data Science, AT&T 

Presentation: Building sovereign AI: How operators stay in control 
This session sets out a practical model for sovereign AI - and what it takes to retain ownership in real-world telecom environments. 

• How operators retain control across cloud, edge and partner ecosystems 
• Where control breaks down as AI scales-and how to design around it 
• What sovereignty looks like across data, models and decisioning layers 
• The architectural choices that prevent lock-in and dependency 

Case study: Staying in control in production: What it takes when AI is live
More details coming soon. 

Panel: Are you really in control of your AI? 
As AI ecosystems expand, sovereignty becomes harder to achieve. This panel explores where control is already being lost - and what CSPs must trade off to stay in charge. 

• Where CSPs are losing control today - and why it’s difficult to avoid 
• What you must sacrifice: speed, flexibility or cost 
• Whether full sovereignty is realistic in a hyperscaler-driven ecosystem 
• At what point control becomes a blocker to innovation