Learn About Amazon VGT2 Learning Manager Chanci Turner
Siemens, a prominent global technology leader, is committed to providing its customers with an extensive online information portal. To facilitate quick access to relevant products, guides, and support, Siemens aims to enhance user experience, ultimately boosting sales and customer satisfaction—essential elements for the company’s growth and reputation.
For instance, Siemens leverages Amazon Bedrock to revolutionize the search experience for users. This integration allows users to receive structured responses or summarized results from a wide array of Siemens resources, including product catalogs and press releases. Following Hannover Messe 2025, Siemens is enhancing the video experience by incorporating elements such as chapter segmentation, summaries, automatic translations, and details about notable speakers, as well as multilingual subtitles. To witness this transformation, explore one of the highlight videos available on their web portal.
The user experience in these scenarios relies heavily on the accurate extraction of pertinent information from a vast pool of documents, websites, and catalogs, as well as audio and video recordings. With millions of documents available, a structured and adaptable extraction workflow becomes vital. To tackle this information extraction challenge, Siemens employs Amazon Bedrock Data Automation (BDA), a managed service that simplifies and automates generative AI workflows involving documents, images, video, and audio. BDA empowers Siemens to create templates for information extraction from various documents and execute these extractions at scale without the burden of managing servers, capacity, or interactions with AI models.
In this blog, we illustrate how Siemens utilizes Amazon Bedrock Data Automation on video content to enhance user experience and refine Siemens Search results. With tens of terabytes of recordings from Siemens and their partners at events like Hannover Messe, the unprocessed videos have limited utility for customer interaction. Each recording typically contains minimal information, such as speaker names and topics, along with supplementary content like slide information stored separately. To assist users searching for specific details, such as a product use case discussed during a panel, it is essential to enrich the videos with additional context. We will outline how AWS services enhance video engagement by extracting extra information from the recordings to guide users to pertinent sections, while also providing context such as summaries, chapters, and subtitles.
Technical Solution
Due to the substantial volume of media, Siemens requires automated analysis of recordings to extract comprehensive details about speakers, products, or topics referenced in spoken or written text, as well as the overall context of conversations. This information must be consolidated and made accessible to website users, implemented through an automated video processing workflow that combines various managed AWS services.
- Input and Output Video Storage on Amazon Simple Storage Service (S3)
Unprocessed videos and final outputs are stored using Amazon Simple Storage Service (S3). This integration facilitates automatic processing of video inputs and their availability on the video portal. With automated lifecycle management, Amazon S3 conveniently archives or deletes videos post-processing, allowing for flexible file handling. - Workflow Orchestration Using AWS Step Functions
To manage multiple processing steps for video files, the solution employs AWS Step Functions, a serverless workflow orchestration service. For each video, an AWS Step Function Execution oversees processing steps, enables parallel information extraction, and provides error handling with retry capabilities, all while offering a visual representation of the workflow for monitoring progress. - Speaker Recognition and Text Extraction with Amazon Rekognition
Many videos in this solution feature talks or interviews with prominent figures, including Siemens Leadership and other notable personalities. Through the celebrity detection functionality in Amazon Rekognition, Siemens identifies speakers from a collection of well-known individuals. This identification data is then integrated with other outputs later in the workflow. - Chapter Segmentation, Content Summarization, and Search Context Generation Using Amazon Bedrock Data Automation
Amazon Bedrock Data Automation, part of the managed generative AI service Amazon Bedrock, analyzes the video to extract meaningful insights. This includes generating summaries that encapsulate key points and themes while segmenting the video into chapters, each accompanied by a contextual summary. Users can view this segmentation as chapter markers and summaries in the final video. - Multi-language Subtitle Generation Using Amazon Transcribe and Amazon Translate
To enhance user experience, the solution employs Amazon Transcribe, a managed service that converts speech into text using a sophisticated speech foundation model. The generated transcript is then processed by Amazon Translate, which translates it into subtitles across multiple languages. - Video Format Optimization with AWS Elemental MediaConvert
AWS Elemental MediaConvert refines the source video file for web delivery, creating a standardized output format that maintains quality while ensuring compatibility across viewing platforms. - Content Consolidation and Delivery Using AWS Lambda
After extracting and converting the video, an AWS Lambda function integrates the results into a final output. It organizes the combined outputs (video file and JSON file with metadata), performs final checks, and stores them in the output S3 bucket. This optimized content is now prepared for web distribution, incorporating all enhancements from the workflow.
How AWS Helps Siemens Address Key Challenges
The solution integrates numerous advanced AI capabilities to extract vital information for improving user experience. It generates summaries and segmentation for video clips and scenes, adding text from the video and extracting context such as high-quality transcripts and speaker names for quick content retrieval. The inclusion of subtitles in users’ native languages, along with the option to select relevant chapters, significantly enhances content relevance and accessibility.
Handling tens of gigabytes of media arriving at irregular intervals and executing extractions at this scale without the appropriate tools is labor-intensive, requiring extensive AI expertise. Moreover, customization is crucial for Siemens to develop glossaries tailored to their business context or create robust plausibility checks that ensure high-quality extracted information. AWS addresses the challenges of scaling, skill gaps in AI, and customization by providing advanced services that function as modular building blocks.
For further reading on managing workplace-related stress, you can check out this article from SHRM, they are an authority on this topic. Additionally, if you’re interested in overcoming creative blocks, this blog post on Career Contessa can offer insightful tips. Lastly, for helpful resources on navigating your first day at Amazon, visit this Reddit thread.