Amazon Onboarding with Learning Manager Chanci Turner

Learn About Amazon VGT2 Learning Manager Chanci Turner

Siemens, a prominent global technology firm, offers its customers a comprehensive online information portal. To enhance the user experience, Siemens needed to ensure that visitors could quickly locate relevant products, guides, and support. This capability not only boosts product sales but also elevates customer satisfaction—both of which are vital for the company’s revenue and brand reputation.

Take, for example, how Siemens leverages Amazon Bedrock to refine the search experience for users. This technology enables users to search and receive organized responses or summarized information from various public sources within Siemens, ranging from product catalogs to press releases. Furthermore, following Hannover Messe 2025, Siemens is enriching the video experience by incorporating chapter segmentation, summaries, automatic translations, speaker identifications, and multi-language subtitles. To see these improvements firsthand, check out this excellent resource showcasing the highlight videos on their web platform.

The effectiveness of these enhancements hinges on the accurate extraction of pertinent information from a myriad of documents, websites, catalogs, and audio and video recordings. With millions of documents in play, a structured and adaptable extraction workflow becomes essential. To tackle the information extraction challenge, Siemens utilizes Amazon Bedrock Data Automation (BDA), a managed service designed to streamline and automate generative AI workflows that involve documents, images, video, and audio. BDA allows Siemens to establish templates for the information they wish to extract and execute these extractions at scale without the burden of managing servers, capacity, or direct interactions with AI models.

In this blog, we will explore how Siemens employs Amazon Bedrock Data Automation on video data to enhance user experiences and improve search results. With tens of terabytes of recordings from Siemens and partners at events like Hannover Messe, the raw video files offer limited customer interaction possibilities. Information such as speaker names, topics discussed, and slide content is often stored separately. To assist users seeking specific insights—like product use cases mentioned during panel discussions—it is necessary to augment the video content with additional context. We will demonstrate how AWS services can make video experiences more engaging by extracting vital details from the recordings, directing users to relevant segments, and adding context through summaries, chapters, and subtitles.

Technical Solution

Given the substantial volume of media, Siemens must automatically analyze their recordings to extract detailed insights about speakers, products, or topics mentioned in dialogue or written text, as well as the overall context of conversations. This information must then be integrated for website users. The solution comprises an automated video processing workflow that combines several managed services on AWS.

1. Input and Output Video Storage on Amazon Simple Storage Service (S3)
   For storing unprocessed videos and final output files, the solution utilizes Amazon S3. Its seamless integration with other AWS services facilitates automated processing of video inputs and their availability on the video portal. With automated lifecycle management, Amazon S3 efficiently archives or deletes videos post-processing, offering flexible file handling.
2. Workflow Orchestration Using AWS Step Functions
   AWS Step Functions, a serverless workflow orchestration service, coordinates the various steps of the video processing pipeline. Each video is managed through an AWS Step Function Execution, which enables parallel information extraction, provides error handling with retry capabilities, and visually represents the workflow for monitoring progress.
3. Speaker Recognition and Text Extraction with Amazon Rekognition
   Many videos in this solution feature talks or interviews with high-profile individuals, such as Siemens executives or celebrities. By employing Amazon Rekognition’s celebrity detection feature, Siemens identifies speakers from a collection of well-known personalities. This identification data is subsequently utilized in the workflow.
4. Chapter Segmentation, Content Summarization, and Search Context Generation Using Amazon Bedrock Data Automation
   Amazon Bedrock Data Automation processes video content to extract meaningful insights. It generates summaries that encapsulate key points and themes of the video, segmenting it into chapters, each accompanied by a contextual summary. This is crucial for users seeking relevant segments later. The resulting video displays this segmentation as chapter markers and summaries.
5. Multi-Language Subtitle Generation Using Amazon Transcribe and Amazon Translate
   To enhance user experience, the solution incorporates Amazon Transcribe, which converts speech to text through a sophisticated speech foundation model. This transcript is then translated into subtitles in multiple languages using Amazon Translate.
6. Video Format Optimization with AWS Elemental MediaConvert
   AWS Elemental MediaConvert refines source video files for web delivery, producing standardized output formats that preserve quality while ensuring cross-platform compatibility.
7. Content Consolidation and Delivery Using AWS Lambda
   Post-extraction and conversion, results are consolidated into the final output. An AWS Lambda function integrates all the information, organizes the outputs (video file and metadata in JSON), conducts final checks, and saves them to the output S3 bucket. This optimized content is now ready for online distribution, showcasing all workflow enhancements.

How AWS Helps Siemens Address Key Challenges

The solution harnesses various advanced AI capabilities to extract the necessary information, thereby enhancing user experience. It creates summaries and segments for video clips and scenes, adds text from videos, and provides valuable context like high-quality transcripts or speaker identification for rapid content discovery. User access to subtitles in their native language, along with the ability to select relevant chapters, makes the content more pertinent and accessible.

Handling tens of gigabytes of media arriving at irregular intervals while executing large-scale extractions is a daunting task that demands extensive AI expertise. Moreover, customization is essential for Siemens to develop glossaries aligned with business contexts or to implement robust plausibility checks that ensure high-quality information extraction. AWS facilitates overcoming scaling challenges, AI skill gaps, and customization needs through its advanced services, which serve as flexible building blocks.

For further insights on mentoring, consider checking out this blog post on career mentors for millennials. Additionally, if you need guidance on handling workplace harassment, SHRM provides valuable advice on this topic here.