The data product challenge - Feeding AI with trusted, industrial-scale data

Chair: Roz Roseboro, Senior Principal Analyst, Omdia

From AI pilots to reinventing how work gets done

• Redesign work from the ground up: How AT&T Business brought business and technology teams together to simplify operations and reimagine work across complex legacy environments and fragmented customer journeys
• Move beyond AI pilots to production: Lessons learned from creating a new operating model, establishing governance, and transforming the workforce to scale AI effectively
• Deliver meaningful customer outcomes: What changed, what worked, and what AT&T Business learned on its journey to simpler, more connected operations.

Fred Brown, Leader of AI for Telecoms, PwC
Jennifer Van Buskirk, Head of Business Operations, AT&T 

Which sites should you NOT upgrade? The next-G capital question network data can't answer.  

• Where will the investment actually pay back? Every decade, operators make a nine- or ten-figure capital investment: upgrade to the next G. As they look towards 6G, the question is not just what the network can deliver, but whether the investment will earn a return. Traffic and density show network demand; they don't show which sites should be funded, deferred, or excluded based on commercial return.
• A real investment decision: VIVA Bolivia’s CEO will show how its next-G evaluation changes when network-planning data is connected with revenue, customer, and cost-to-serve evidence.
• Giving AI the context to make the call: Totogi’s CEO will explain how a governed telco ontology, aligned with TM Forum ODA and SID and deployed as an overlay on existing systems, lets AI evaluate investment at site level. Attendees will leave with a checklist for their own next-G build.

Danielle Rios, CEO, Totogi 
Ryan Alvarez, CEO, VIVA Bolivia 

Giving AI a common language

• AI agents struggle to make reliable network decisions when operational systems describe the same infrastructure using different identities, definitions and relationships
• Google Cloud and NumoData explore how shared, governed semantics give agents a consistent understanding of network objects and their dependencies
• This common understanding allows agents to collaborate across systems, platforms and technology domains without losing the meaning of the underlying network.

Robin Osagie, Ontology Strategy Lead, NumoData
Piyush Mathur, Group Product Manager, Google Cloud

OTel: Hardware and software agnostic training and inference from data centers to the edge

• Train and deploy AI anywhere: Hardware- and software-agnostic training and inference, from data centers to edge devices
• Scale without lock-in: Training on 440B tokens with portable workflows spanning AMD, NVIDIA, Apple, and Intel hardware
• Build open, portable ecosystems: Lessons from creating and releasing Open Telco (OTel) as an open-source ecosystem, with an emphasis on portability and interoperability

Farbod Tavakkoli, Data Scientist, AT&T

Panel: Fix the data layer or fail at AI: What needs to change now? 
Despite major investment in AI, many operators still struggle to scale due to fragmented, inconsistent, and untrusted data. This panel explores what is really holding CSPs back - and what it takes to build data foundations that support AI-native operations at scale. 

• What ‘trusted data’ looks like in practice 
• Where fragmentation still breaks cross-domain orchestration 
• Who owns the data – and who is accountable for quality 
• What changes when data becomes a product, not an asset 

Satya Malkapurapu, Senior Data Engineering Executive, Verizon 
Bo Gu, Director of Data Science, AT&T
Rick Lievano, CTO, Telecommunications, Microsoft 
Robin Osagie, Ontology Strategy Lead, NumoData