Team Global Express deploys 12 AI agents in logistics operations

Team Global Express, a logistics provider spun out of Toll Group in 2021, has identified 73 AI use cases and says it has moved 12 "AI agents" into production with a further five proofs-of-concept, Michael Farrar told the AWS Summit Sydney, according to ITNews. Farrar described centralising previously siloed line-of-business data into a unified platform on AWS, adopting Redshift for analytics and DynamoDB for operational stores, and progressively adding metadata and security guardrails to improve data quality and queryability, per the ITNews report. A Salesforce customer case study and event coverage document deployment of an agent called Āwhina via Salesforce Agentforce into track-and-trace workflows, with Salesforce reporting a 25% reduction in call volume within weeks and Team Global Express estimating a 50% reduction in routine inquiries in the published case study.

Per the ITNews account of Farrar's remarks, Team Global Express uses Amazon Rekognition to analyse proof-of-delivery images and remove visible personally identifiable information such as house numbers and parcel labels. The company is also building operational intelligence for frontline staff using Amazon Bedrock AgentCore, as described in Farrar's presentation. The Salesforce customer story states that Agentforce integrates directly with the carrier's track-and-trace system via a real-time API to answer delivery-status queries and escalate complex cases to human agents.

Editorial analysis: Companies that have consolidated scattered operational data into centralised cloud stores frequently find they can move from experimentation to production faster, because consistent schemas, metadata, and guarded access reduce friction when connecting models to operational systems. Industry reporting from McKinsey and vendor case studies have highlighted the same pattern where agentic workflows yield measurable contact-center relief and throughput gains when connected to reliable track-and-trace data.

Editorial analysis: For practitioners, the story is notable for two reasons. First, the scale of planned use cases-73 identified opportunities with 12 live agents-illustrates the breadth of potential agent applications in logistics, from PII handling to frontline decision support. Second, the stack described (cloud data warehouse, operational NoSQL store, image analysis, and agent orchestration) is a representative reference architecture for productionising agentic features in customer-facing and operational workflows. The Salesforce numbers on call-volume reduction provide a concrete operational KPI that teams can benchmark against, while vendor-provided results should be treated as indicative rather than independently verified.

Editorial analysis: Observers should track a few signals to assess how replicable these results are across logistics operators: the extent of data standardisation across lines of business, measured improvements in contact-center KPIs beyond initial weeks, the set of edge cases that still require human escalation, and how PII-removal models affect downstream audit and compliance workflows. Also watch for third-party partnerships and platform choices-Agentforce, Amazon Bedrock AgentCore, Rekognition, Redshift, and DynamoDB-as they shape integration and operational cost trade offs.