Intuit delivers explainable tax insights with Amazon Bedrock and Anthropic Claude, integrated with TurboTax

Intuit uses Amazon Bedrock and Anthropic Claude with its proprietary tax engine to generate contextual explanations for tax calculations for TurboTax customers. The solution is integrated with Amazon Connect Customer for contact center support and Amazon SageMaker for custom model work in production.

Organization
Intuit
Industry
Finance
Published
May 2026

Reported outcomes

Strategic outcomes

Customer experience & trustImproved customer experience across channelsScale & capacitySupported large-scale personalizationNew product / capabilityGenerated contextual tax explanationsCustomer experience & trustEnabled integrated contact center support

Primary read

Use case focus

Showing 3 of 4

  • 1Customer Support
  • 2Tax Preparation
  • 3Generative AI
Intuit needed to provide easy-to-understand, contextual explanations for tax calculations at scale while maintaining real-time accuracy checks for millions of tax filers.
  • Intuit built a generative AI solution in Amazon Bedrock using Anthropic Claude.
  • The solution is underpinned by Intuit’s proprietary tax engine to generate explanations with real-time accuracy checks.
  • Intuit also uses Amazon Connect Customer for an integrated contact center and Amazon SageMaker to customize foundation models for production use.
  • The solution benefits millions of TurboTax customers during the 2024 tax season.
  • It improves customer experience across contact channels.
  • It supports large-scale personalization.
Architecture

A generative AI solution inside TurboTax built on Amazon Bedrock with Anthropic Claude, connected to Intuit’s proprietary tax engine for real-time accuracy checks, alongside Amazon Connect Customer and Amazon SageMaker for support and model customization.

Sources & evidence1
Groundedness: 3/5Type: Customer StoryPublished: May 13, 2026Publisher: AWSEvidence: PrimaryConfidence: High

AI-generated summary. Verify important details with the linked sources before relying on this case.

Explore related AI use cases

Was this useful?