---
title: "How TUI scales threat modeling for 100+ teams with Amazon Bedrock"
description: "TUI’s four-person security team faced a bottleneck: manual threat modeling for 100+ digital product teams, each assessment taking 4-6 hours. They built a system using Amazon Bedrock with Claude Sonnet 4.5 and deployed on Amazon Bedrock AgentCore runtime after a 10-week proof of concept with AWS. Instead of loading all 192 security standards into prompts, they implemented metadata-driven selection, reducing context from 338 KB to 21 KB and improving consistency. The LangGraph-based agentic workflow uses self-correcting threat generation and SME comparison for accuracy and completeness. In production, assessment time dropped by up to 70%, sessions per month increased from 3-4 to 10-12, and 255 users registered in eight weeks, making threat modeling mandatory for all new services."
---

# How TUI scales threat modeling for 100+ teams with Amazon Bedrock

[AWS](https://yomu.fyi/company/aws) · Vushesh Babu Adhikari · Nov 5, 2026

**Type:** Explainer

## Summary

TUI’s four-person security team faced a bottleneck: manual threat modeling for 100+ digital product teams, each assessment taking 4-6 hours. They built a system using Amazon Bedrock with Claude Sonnet 4.5 and deployed on Amazon Bedrock AgentCore runtime after a 10-week proof of concept with AWS. Instead of loading all 192 security standards into prompts, they implemented metadata-driven selection, reducing context from 338 KB to 21 KB and improving consistency. The LangGraph-based agentic workflow uses self-correcting threat generation and SME comparison for accuracy and completeness. In production, assessment time dropped by up to 70%, sessions per month increased from 3-4 to 10-12, and 255 users registered in eight weeks, making threat modeling mandatory for all new services.

## Context

TUI’s Platform & Engineering division, with only four security engineers, had to support over 100 digital product teams. Each threat modeling session manually took 4-6 hours, limiting the team to 3-4 sessions per month and creating a constant backlog. The team needed threat modeling to be fast enough to become mandatory for every new service.

## Approach / What changed

TUI collaborated with the AWS Generative AI Innovation Center on a 10-week PoC to build an AI-powered threat modeling system. The solution uses Amazon Bedrock with Claude Sonnet 4.5, metadata-driven selection of security standards, and a LangGraph-based agentic workflow on Amazon Bedrock AgentCore runtime. The workflow processes architecture diagrams through specialized nodes for component identification, threat generation, gap analysis, and reporting.

## Takeaways

- Metadata-driven selection reduced the context from 338 KB to 21 KB, a 94% reduction, by pre-generating keyword and category metadata for each standard.
- The system uses a self-correcting threat generation workflow with iterative refinement and SME comparison for consistency.
- After deployment, assessment time dropped by up to 70%, sessions per month increased from 3-4 to 10-12, and standards were applied 100% consistently.

## FAQ

### How did TUI handle the large number of security standards?

TUI initially tried loading all 192 standards into the prompt, which failed. Instead, they pre-generated metadata for each standard and used metadata-driven selection to pick only the 16 most relevant ones, reducing context from 338 KB to 21 KB.

### What AWS services were used in the solution?

The solution used Amazon Bedrock (Claude Sonnet 4.5), Amazon Bedrock AgentCore runtime, AWS Lambda, Amazon DynamoDB, Amazon S3, AWS Amplify, and Amazon Cognito.

### What results did TUI achieve after deployment?

Assessment time reduced by up to 70%, sessions per month increased from 3-4 to 10-12, and standards were applied consistently. Within four weeks, 64 users registered and produced 34 threat models.

**Tags:** [Agentic AI](https://yomu.fyi/topic/agentic-ai), [Amazon Bedrock](https://yomu.fyi/topic/amazon-bedrock), [Generative AI](https://yomu.fyi/topic/generative-ai), [Security](https://yomu.fyi/topic/security), [Threat Modeling](https://yomu.fyi/topic/threat-modeling)

- Source: [AWS](https://aws.amazon.com/blogs/architecture/how-tui-scales-threat-modeling-for-100-teams-with-amazon-bedrock)
- Source URL: https://aws.amazon.com/blogs/architecture/how-tui-scales-threat-modeling-for-100-teams-with-amazon-bedrock
- Ingested by Yomu: 2026-11-05T21:00:13.578Z

[Read original post](https://aws.amazon.com/blogs/architecture/how-tui-scales-threat-modeling-for-100-teams-with-amazon-bedrock)
