What challenges did DoorDash face before AI implementation?

DoorDash handled a high volume of support calls from Dashers and needed a scalable self-service experience. The experience had to be fast and reliable with minimal latency because Dashers prefer voice support while on the road.

What was the impact of the AI solution?

The solution responds to Dashers in 2.5 seconds or less. It fields hundreds of thousands of calls per day. It was built in about 2 months.

Source

ExpandedEvidence: Low35/100

DoorDash built low-latency voice self-service generative AI agents for Dashers using Amazon Connect, Amazon Lex, and Amazon Bedrock Knowledge Bases

Use case typeCustomer support automationUpdated Jun 13, 2026

DoorDash built a low-latency voice self-service experience for Dashers using Amazon Connect, Amazon Lex, Amazon Bedrock, and Amazon Bedrock Knowledge Bases. The solution was developed with AWS Generative AI Innovation Center to reduce live-agent burden while maintaining issue-resolution quality.

Organization: DoorDash
Industry: Retail
Location: United States
Published: September 2024

Reported outcomes

100,000 calls per day

call volume handledOther quantified impact

2.5 secondsresponse latency2 monthsimplementation time

Strategic outcomes

Speed & agilityBuilt low-latency voice self-serviceCost efficiencyReduced live-agent burdenNew product / capabilityAdded voice-enabled self-service agent

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

Normalized claim

Response latency: 2.5 seconds

AWS Machine Learning BlogSep 24, 2024Blog postExplicit claimLow evidence strength

responding to Dashers with answers to their questions in 2.5 seconds or less

Normalized claim

Call volume handled: 100,000 calls per day

AWS Machine Learning BlogSep 24, 2024Blog postExplicit claimLow evidence strength

The solution fields hundreds of thousands of calls per day

Normalized claim

Implementation time: 2 months

AWS Machine Learning BlogSep 24, 2024Blog postExplicit claimLow evidence strength

DoorDash built a solution ... in just 2 months

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

Customer: DoorDash
Provider: AWS
Maturity: Unknown
Linked source: AWS Machine Learning Blog

No explicit deployment-stage evidence found.

Customer identity supportedSource describes one deploymentMaturity evidence evaluated

Primary read

Use case focus

Showing 3 of 4

1Customer support automation
2Conversational AI
3Retrieval-augmented generation

DoorDash handled a high volume of support calls from Dashers and needed a scalable self-service experience.
The experience had to be fast and reliable with minimal latency because Dashers prefer voice support while on the road.

DoorDash built a voice-enabled RAG self-service agent with Amazon Connect, Amazon Lex, Amazon Bedrock, Amazon Bedrock Knowledge Bases, and Amazon OpenSearch Serverless.
The solution includes automated testing, conversation analytics, monitoring, observability, and hallucination prevention and detection.

Technologies

Amazon Connect Amazon Lex Amazon Bedrock Amazon Bedrock Knowledge Bases Amazon OpenSearch Serverless AWS Lambda Amazon S3 Amazon Data Firehose AWS Glue Amazon QuickSight Amazon SageMaker

The solution responds to Dashers in 2.5 seconds or less.
It fields hundreds of thousands of calls per day.
It was built in about 2 months.

Architecture

DoorDash and the AWS Generative AI Innovation Center built a voice-enabled self-service contact-center solution using Amazon Connect for telephony, Amazon Lex for conversational routing, Amazon Bedrock and Bedrock Knowledge Bases for RAG-based answers, and Amazon OpenSearch Serverless as the retrieval store. The article also describes optional conversation analytics, hallucination detection, and automated testing components.

Sources & evidence1

Evidence: Low35/100Evidence strength