CVPR 2023 in Vancouver highlighted the rapid advancements in computer vision and AI, drawing over ten thousand attendees from related fields and showcasing more than two thousand research papers. A key theme was the repurposing of established computer science solutions, such as using techniques from computer graphics and visual effects in computer vision, exemplifying how procedural models and geometry extraction processes can be adapted to AI. Additionally, innovations in embeddings, like HierVL and visual DNA, are making data exploration and understanding more efficient, allowing AI developers to improve model accuracy and performance by focusing on significant dataset features. The conference also emphasized the role of foundation models, such as OpenAI's CLIP and Meta’s Segment Anything Model, in transforming AI development by bridging textual and visual data and enabling new applications like semantic segmentation and zero-shot diagnosis. The evolving challenge is integrating these models with existing MLOps infrastructure to enhance processes like A/B testing and data enrichment, a need addressed by Labelbox's new Model Foundry solution.