Heathrow Deploys Salesforce AI Assistant Hallie for Passengers

Salesforce announced that Heathrow Airport has deployed an AI service agent called Hallie using Salesforce's Agentforce platform, according to a Salesforce press release. The press release reports Hallie resolves 90% of chat queries without human transfer and draws on Heathrow's Service Cloud knowledge base, live APIs, maps, and flight-status data. Salesforce materials also cite expected gains including up to 40% improvement in digital contact efficiency and more than 50% fewer chat exchanges to reach answers.

Per Salesforce documentation and customer case materials, Hallie is implemented via the Agentforce agent authoring and runtime stack and is accessible to travelers through WhatsApp; Salesforce says the agent uses indexed customer and operational data to provide instant answers such as gate directions, amenity locations, and wait-time information. TechTarget and AI Business coverage place this deployment in the broader Agentforce 360 product roadmap, which introduces hybrid reasoning, voice interfaces, and context indexing to support agentic workflows.

Business Insider reports that Peter Burns, Heathrow's director of digital, e-commerce, and marketing, said Heathrow first tested Salesforce's generative and agentic AI products in 2023 and launched Hallie in March 2025, per Business Insider. Business Insider adds that Heathrow is looking to extend Hallie to its website and app later in the year. Salesforce's customer story materials reiterate Heathrow's scale, noting the airport handles over 83 million passengers annually.

Industry observers: Deployments that embed agentic AI into customer-facing channels typically rely on an integrated data foundation to reach high resolution rates; Salesforce's materials highlight that Heathrow consolidated CRM, loyalty data, Wi‑Fi signals, and bookings into a single knowledge layer to feed Hallie. Comparable enterprise implementations emphasize the difference between a high-performing assistant and a brittle pilot is often data integration and content curation rather than model novelty.

Industry coverage from TechTarget and AI Business frames this as part of Salesforce's broader push to make Slack and its CRM the interface for agentic assistants, with Agentforce 360 adding hybrid reasoning and voice features. Analysts quoted in that coverage argue that the more customers normalize and integrate enterprise data, the more agentic systems can solve complex, multi-step tasks rather than only answering FAQs.

Observed patterns in similar transitions: High reported resolution rates, like the 90% figure Salesforce cites, commonly follow significant manual effort to author, tag, and maintain knowledge artifacts; Salesforce's customer story notes Heathrow uploaded roughly 800 knowledge articles for Hallie's training. Practitioners should treat headline metrics as dependent variables of data quality, content engineering, and monitoring pipelines rather than as model performance alone.

Reporting to follow will include independent evaluations of operational impact (call-volume reduction, average handling time, customer satisfaction) and any public metrics on error rates or escalations. Observers should also watch whether Heathrow or third parties publish post-deployment telemetry or user-experience studies that corroborate vendor-reported gains.

Salesforce and Heathrow present Hallie as an operationalized agentic-AI deployment with substantial vendor-reported efficiency gains. Editorial analysis: Companies deploying agentic assistants at scale commonly pair them with disciplined data integration and content workflows; the Heathrow case provides a useful reference point for practitioners instrumenting enterprise agents for high-volume, time-sensitive use cases.