What challenges did Gathern face before AI implementation?

The monolithic architecture could not scale safely as demand grew rapidly. The existing filter-based search system could not interpret guests' natural-language requests in Arabic and English.

Source

Evidence: Medium50/100

Gathern: AI-powered natural language search using Amazon Bedrock and modernization on Amazon EKS

Use case typeSearch modernizationUpdated Jun 8, 2026

Gathern modernized its infrastructure on AWS by moving from a monolithic system to a microservices-based architecture. It added Amazon Bedrock-powered natural language search so travelers could describe accommodation needs in Arabic or English and have them converted into structured search filters.

Organization: Gathern
Industry: Other
Location: Saudi Arabia
Published: June 2026

Reported outcomes

90,000 requests per minute

peak traffic handledRisk, reliability & safety

−77%average API latency87 releasesreleases in a single month

Strategic outcomes

Speed & agilityContinuous deployment replaced monthly releasesCustomer experience & trustNatural-language search in Arabic and EnglishScale & capacityA foundation for continued innovation and growth

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

Normalized claim

Average API latency: 77% decrease

AWS Solutions Case StudiesJun 8, 2026Customer storyExplicit claimMedium evidence strength

decreased its average API latency from 600 milliseconds to 140 milliseconds, a reduction of approximately 77 percent

Normalized claim

Peak traffic handled: 90,000 requests per minute increase

AWS Solutions Case StudiesJun 8, 2026Customer storyExplicit claimMedium evidence strength

the system handled peak traffic of 90,000 requests per minute without incident

Normalized claim

Releases in a single month: 87 releases increase

AWS Solutions Case StudiesJun 8, 2026Customer storyExplicit claimMedium evidence strength

We have been able to release 87 times in a single month

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

Customer: Gathern
Provider: AWS
Maturity: Unknown
Linked source: AWS Solutions Case Studies

No explicit deployment-stage evidence found.

Customer identity supportedSource describes one deploymentMaturity evidence evaluated

Primary read

Use case focus

Showing 2 of 2

1Search modernization
2Workflow automation

The monolithic architecture could not scale safely as demand grew rapidly.
The existing filter-based search system could not interpret guests' natural-language requests in Arabic and English.

Built a microservices architecture on Amazon EKS with supporting AWS services such as Amazon SQS, Amazon Aurora and AWS Graviton processors.
Used Amazon Bedrock to build natural-language search and Amazon Rekognition for automated image moderation.

Technologies

Amazon Bedrock Amazon EKS AWS Graviton Processors Amazon SQS Amazon Aurora Amazon Rekognition

Average API latency fell from 600ms to 140ms.
The platform handled peak traffic of 90,000 requests per minute without incident.
Release frequency increased to 87 times in a single month.

Architecture

Gathern rebuilt its platform on Amazon EKS in a microservices architecture, using Amazon SQS for asynchronous messaging, Amazon Aurora for the database layer, AWS Graviton processors for performance and cost optimization, Amazon Bedrock for natural-language search, and Amazon Rekognition for image moderation.

Sources & evidence1

Evidence: Medium50/100Evidence strength