What challenges did Grupo Falabella face before AI implementation?

Standardize and accelerate incident reporting and ticket creation across the organization. Reduce delays from manual or informal channels and improve ticket categorization and downstream resolution efficiency. Improve internal support experience and reduce operational friction in a growing regional retail business.

Source

GCPProductionEvidence: Medium65/100

Falabella — TARS AI agent for incident/ticket creation and deflection (Dialogflow + Contact Center AI + Vertex AI Search)

Use case typeIT operationsUpdated Jun 13, 2026

Grupo Falabella (Falabella Retail / Sodimac / Tottus) built TARS, a multi-agent conversational AI workflow for incident reporting and ticket creation across retail and help desk operations in Latin America. The solution uses Google Cloud Contact Center AI Platform, Dialogflow/Conversational Agents, Gemini models, Vertex AI Search for RAG over internal process documentation, and supporting services including BigQuery, Pub/Sub, Cloud Run, Firestore, Cloud Storage, Dataform, Looker, and Cloud Logging. The system captures key incident fields, auto-creates and routes tickets, centralizes communication across channels, and supports regional standardization of support processes.

Organization: Grupo Falabella
Industry: Retail
Location: Chile
Published: June 2026

Reported outcomes

40-80%

accuracyQuality & accuracy

16 minutestime4 minutestime2 hourstime30 minutestime−33%quantified impact

Strategic outcomes

New product / capabilityBuilt automated incident ticketing workflowSpeed & agilityStandardized and accelerated incident reportingCustomer experience & trustImproved internal support experienceBetter decisions & insightImproved ticket categorization accuracy

Catalog median for quality & accuracy deployments: +41% across 63 reported metrics. Compare benchmarks →

Why do we believe this?Outcome claims, sources, and evidence checks

Normalized claim

Time: 16 minutes decrease

Google Cloud Customer StoriesJun 3, 2026Customer storyInferred claimMedium evidence strength

Ticket creation time reduced from 16 minutes to 4 minutes.

Normalized claim

Time: 4 minutes decrease

Google Cloud Customer StoriesJun 3, 2026Customer storyInferred claimMedium evidence strength

Ticket creation time reduced from 16 minutes to 4 minutes.

Normalized claim

Time: 2 hours decrease

Google Cloud Customer StoriesJun 3, 2026Customer storyInferred claimMedium evidence strength

Ticket resolution time reduced from about 2 hours to 30 minutes.

Normalized claim

Time: 30 minutes decrease

Google Cloud Customer StoriesJun 3, 2026Customer storyInferred claimMedium evidence strength

Ticket resolution time reduced from about 2 hours to 30 minutes.

Normalized claim

Accuracy: 40-80% increase

Google Cloud Customer StoriesJun 3, 2026Customer storyInferred claimMedium evidence strength

Ticket categorization accuracy improved from about 40% to 80%.

Normalized claim

Quantified impact: 33% decrease

Google Cloud Customer StoriesJun 3, 2026Customer storyInferred claimMedium evidence strength

Human-agent ticket creation reduced by 33%.

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: Grupo Falabella, Falabella Retail, Sodimac, Tottus
Provider: GCP
Maturity: Production
Linked source: Google Cloud Customer Stories

Improve internal support experience and reduce operational friction in a growing regional retail business

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 3 of 4

1IT Service Management
2Incident Management
3Workflow Automation

Standardize and accelerate incident reporting and ticket creation across the organization.
Reduce delays from manual or informal channels and improve ticket categorization and downstream resolution efficiency.
Improve internal support experience and reduce operational friction in a growing regional retail business.

Built TARS as a multi-agent conversational workflow to capture incident details, retrieve context with Vertex AI Search over internal process documentation, categorize incidents, and create and assign tickets automatically.
Used Contact Center AI Platform and Dialogflow as the main conversational layer, with Gemini models to enable natural interactions.
Integrated Pub/Sub, Cloud Run, BigQuery, Firestore, Cloud Storage, Dataform, Looker, and Cloud Logging to orchestrate events, store fast-access data, centralize analytics, manage content, and improve observability.

Technologies

Google Cloud Contact Center AI Platform Dialogflow Gemini Vertex AI Search BigQuery Pub/Sub Cloud Run Firestore Cloud Logging Cloud Storage Dataform Looker

More than 22,000 tickets created through TARS.
Ticket creation time reduced from 16 minutes to 4 minutes.
Ticket resolution time reduced from about 2 hours to 30 minutes.
Ticket categorization accuracy improved from about 40% to 80%.
Human-agent ticket creation reduced by 33%.

Architecture

TARS is a multi-agent conversational workflow on Google Cloud. It uses Contact Center AI Platform and Dialogflow/Conversational Agents as the conversational front end, Vertex AI Search for RAG over internal process documentation, Pub/Sub for event orchestration, Cloud Run for API integrations and data-loading tasks, BigQuery for analytics, Firestore for fast-access storage, Cloud Storage for documentation, Dataform for data ingestion orchestration, Looker for reporting, and Cloud Logging for observability and alerts. Gemini models power more natural interactions, and the system centralizes incident reporting across multiple channels and regions.

Sources & evidence1

Evidence: Medium65/100Evidence strength