What challenges did tiket.com face before AI implementation?

Provide faster, more seamless customer service for itinerary changes, urgent refunds, and booking support requests at scale. Integrate real-time operational data while improving responsiveness and reducing customer service workload.

What was the impact of the AI solution?

CRATER handles more than 75,000 customer interactions per month versus about 10,000 previously. The solution improved customer satisfaction to 84.6% CSAT. It reduced workload for customer service teams and increased agent productivity.

Source

MicrosoftScaled productionEvidence: Medium65/100

tiket.com: agentic AI assistant for 75,000+ monthly customer interactions using Azure OpenAI & Microsoft Foundry (CRATER)

Use case typeCustomer service automationUpdated Jul 14, 2026

tiket.com, an Indonesian travel and OTA company, built an in-house agentic AI assistant called CRATER on Microsoft Foundry, Azure OpenAI Service, Microsoft AutoGen, and Foundry AI Guardrails to streamline customer service for bookings, refunds, itinerary changes, and related travel requests. The solution uses multi-agent orchestration, selects models per use case, connects to real-time operational data, and is embedded into a single in-app conversation flow to improve service quality, reliability, and cost efficiency.

Organization: tiket.com
Industry: Consumer & Food
Location: Indonesia
Published: July 2026

Reported outcomes

75,000 interactions/month

monthly customer interactions handledOther quantified impact

10,000 interactions/monthmonthly customer interactions previously+84.6%customer satisfaction (CSAT)

Strategic outcomes

Customer experience & trustFaster, more responsive customer serviceCost efficiencyReduced workload for customer service teamsScale & capacityExpanded service capacity at significantly greater scale

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

Normalized claim

Monthly customer interactions handled: 75,000 interactions/month increase

Microsoft SourceJul 14, 2026News articleExplicit claimMedium evidence strength

This AI-powered travel assistant now handles more than 75,000 customer interactions every month

Normalized claim

Monthly customer interactions previously: 10,000 interactions/month increase

Microsoft SourceJul 14, 2026News articleExplicit claimMedium evidence strength

increase from approximately 10,000 monthly interactions previously

Normalized claim

Customer satisfaction (CSAT): 84.6% increase

Microsoft SourceJul 14, 2026News articleExplicit claimMedium evidence strength

high customer satisfaction (CSAT) score of 84.6%

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

Customer: tiket.com
Provider: Microsoft
Maturity: Scaled Production
Linked source: Microsoft Source

Provide faster, more seamless customer service for itinerary changes, urgent refunds, and booking support requests at scale

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 3 of 3

1Customer service automation
2Agent orchestration
3Workflow automation

Provide faster, more seamless customer service for itinerary changes, urgent refunds, and booking support requests at scale.
Integrate real-time operational data while improving responsiveness and reducing customer service workload.

Built CRATER, an in-house agentic AI assistant on Microsoft Foundry and Azure OpenAI Service.
Used Microsoft AutoGen for multi-agent orchestration and Foundry AI Guardrails for safety policies.
Connected the assistant to bookings, payments, and refunds so it can turn customer intent into system actions inside the app.

Technologies

Microsoft Azure Azure OpenAI Microsoft Foundry Microsoft AutoGen Foundry AI Guardrails

CRATER handles more than 75,000 customer interactions per month versus about 10,000 previously.
The solution improved customer satisfaction to 84.6% CSAT.
It reduced workload for customer service teams and increased agent productivity.

Architecture

CRATER is an in-house agentic AI assistant developed on Microsoft Foundry. It uses Microsoft Foundry model catalog to select the most appropriate LLM per agent and use case, Microsoft AutoGen for multi-agent orchestration, and Foundry AI Guardrails to enforce safety policies. The assistant is embedded in a single in-app conversation flow and connects to real-time operational data including bookings, payments, and refunds to translate intent into system actions.

Sources & evidence1

Evidence: Medium65/100Evidence strength