Pixieset, an all-in-one photography business service hosting over 8 billion photos, deployed an AI-generated alt-text feature using Amazon Bedrock that achieved 35 percent adoption among applicable users sixteen months after launch (AWS Machine Learning Blog). The feature moved from concept to production in four months and generated alt text for over 750,000 photos in the first week alone (AWS Machine Learning Blog). Pixieset built the feature on Amazon Bedrock's managed inference, requiring only a single API call without new infrastructure or GPU provisioning, and used Anthropic Claude 3.5 Sonnet for multimodal image understanding and cost efficiency at scale (AWS Machine Learning Blog).
For commerce practitioners, Pixieset's approach offers a critical lesson: AI adoption succeeds when it solves work that users are deferring rather than automating tasks central to their craft (AWS Machine Learning Blog). The feature lets users review AI-generated captions one image at a time before choosing to enable auto-apply, building trust incrementally rather than demanding full automation upfront (AWS Machine Learning Blog). The feature achieved zero downtime since launch by using cross-region inference in Amazon Bedrock and retrying with a secondary model on failures (AWS Machine Learning Blog).