Telstra Plans AI Multi-Agent Systems to Transform Operations

Telstra is setting the stage to integrate multi-agent AI systems into its operations, leveraging a joint venture with Accenture and modernizing its data infrastructure. The telco's Executive Director of Data and AI, Dayle Stevens, revealed plans to "reinvent all of our processes with agentic AI," starting with two initial processes yet to be finalized.

Key Initiatives

• Agentic Architecture: Telstra is prioritizing the development of a platform to orchestrate AI agents from various sources, including existing vendors like Workday, SAP, ServiceNow, and Salesforce, as well as Accenture's AI refinery.
• Silicon Valley Hub: A collaborative space at Accenture's Mountain View office will host experts from partners like Microsoft, Databricks, and AWS to co-design solutions. However, Telstra's team will primarily engage virtually.
• Data Modernization: Telstra is streamlining its data landscape, reducing from 80 to three core platforms (Azure, AWS, and on-premises). A data mesh architecture and Microsoft Fabric are part of this effort.

Strategic Pillars

1. Data Foundations: Focus on accessibility, quality, and protection of data, including network data stored on-premises.
2. Responsible AI: Enhancing real-time model monitoring to align with the rapid adoption of agentic AI.
3. AI Foundations: Building reusable AI components, such as the agentic platform, to avoid case-by-case scaling.

Reuse and Acceleration

Telstra highlights Ask Telstra, an internal AI tool for knowledge base searches, as a success story. Its successor, Ask InfraCo, was deployed 88% faster, demonstrating the benefits of reusable AI frameworks. Stevens emphasized the goal of extending this efficiency to agentic AI.

"We want to move away from scaling AI use case by use case," Stevens said, underscoring the importance of foundational work to unlock broader AI potential.