Telstra Deploys Agentic AI to Detect Migration Errors
Telstra is using agentic AI to inspect customer records after migrations from legacy CRM systems, aiming to reduce call centre volume and speed project timelines. The carrier built agents using `Microsoft AI Foundry` agentic capabilities to reason over migrated data, surface discrepancies, and identify issues before customers call. Telstra frames the project as both an operational relief for frontline staff and a way to create reusable data products that accelerate future system modernisation. Kim Bennemann, Telstra AI solutions group owner, says the approach replaces custom one-off migration tools with smaller, decoupled agent components that perform reasoning across datasets, improving customer experience and reducing migration time.
What happened
Telstra is deploying agentic AI to scan customer records after migrations from legacy relationship management systems, using `Microsoft AI Foundry` to build agents that reason over migrated data and surface discrepancies before customers call. Kim Bennemann, Telstra AI solutions group owner, said the capability reduces pressure on frontline call centre staff, improves customer experience, and shortens migration project timelines.
Technical details
The implementation centers on lightweight, decoupled agents that perform cross-record reasoning rather than bespoke, migration-specific scripts. These agents are constructed with `Microsoft AI Foundry` agentic capability and operate over newly created data products-reusable datasets created during the project. Typical agent functions include:
- •discrepancy detection across migrated records
Agents combine programmatic checks with reasoning to detect issues and perform various functions that are not specific to a single migration use case. That lets Telstra migrate customers faster while catching the types of discrepancies that would otherwise generate reactive call volumes.
Context and significance
This is a pragmatic, production-focused application of agentic AI inside a large telco. It shifts migration tooling from bespoke ETL and manual QA toward reusable, reasoning-driven components and data products. For practitioners, the work demonstrates two trends: the operationalisation of agentic patterns for data-quality workflows, and the emergence of data products as an integration surface between LLM-driven agents and traditional engineering pipelines. The approach reduces operational burden and lowers customer experience risk during large system modernisations.
What to watch
Observe how Telstra measures call-volume reduction and migration velocity, and whether the agents require ongoing human-in-the-loop feedback to maintain precision. Also watch for export of the same agentic pattern to other modernization projects and the governance controls Telstra applies to agent decisions.
Scoring Rationale
This is a notable, practical deployment of agentic AI inside a major telco with clear operational benefits for customer experience and migration velocity. It is not a frontier model or major platform launch, but it demonstrates an actionable pattern for production data-quality automation.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
