What challenges did the organization face before AI implementation?

Sales teams faced difficulties in quickly accessing comprehensive, up-to-date customer data spread across multiple internal and external systems.

Source

Evidence: Low20/100

AWS boosts sales pipeline using generative AI solution built on Amazon Bedrock

Use case typeLegal workflow automationUpdated Jun 13, 2026

AWS Sales and Marketing teams transformed their sales workflows with generative AI to automate repetitive tasks and provide personalized insights, improving sales pipeline management.

Industry: Tech & Comms
Location: United States
Published: August 2024

Reported outcomes

Impact: +4.9%

Other quantified impact

Time: More than 35 minutes

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

Normalized claim

Time: 35 minutes decrease

AWS Machine Learning BlogAug 26, 2024Blog postInferred claimLow evidence strength

Generated over 100,000 GenAI Account Summaries, saving an average of 35 minutes per summary.

Normalized claim

Quantified impact: 4.9% increase

AWS Machine Learning BlogAug 26, 2024Blog postInferred claimLow evidence strength

Sellers reported a 4.9% increase in opportunity value and better preparedness for customer engagements.

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

Customer: Not established
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 2 of 2

1Sales Workflow Automation
2Generative AI for CRM

AWS developed an AI-powered Account Summaries feature integrated with CRM, using Amazon Bedrock and Amazon Q Business to generate personalized account narratives from diverse structured and unstructured data.
The solution uses multiple models including Amazon Titan and Anthropic Claude on Amazon Bedrock, combining strengths for summary generation.
Architecture includes asynchronous processing, modular multi-model selection, robust data indexing, retrieval, and hallucination mitigation strategies.
Security is enforced with row-level access controls and content deletion after delivery.

Technologies

Amazon Bedrock Amazon Q Business Amazon Titan Anthropic Claude models Amazon RDS Amazon S3

Significant positive impact on teams handling multiple accounts and during account transitions.

Architecture

The system uses asynchronous Lambda functions to handle summarization requests, combining multiple LLM models including Amazon Titan and Anthropic Claude on Amazon Bedrock. It indexes and retrieves diverse data sources using vector similarity and cross-encoder models for RAG-enhanced LLM prompting, with rigorous hallucination mitigation strategies.

Sources & evidence1

Evidence: Low20/100Evidence strength