WEBINAR: Embeddings in Practice unlocking New AI Capabilities

Explore satellite data with AI. Detect change, uncover patterns and search imagery using text. In this webinar, Marcin Kluczek, Bartłomiej Ostrowski and Bartosz Augustyn explore how artificial intelligence is opening new ways to work with Earth Observation (EO) data. From Sentinel-1 and Sentinel-2 to embeddings and foundation models, the speakers connect the technology with practical applications: detecting wildfires and other anomalies, mapping environmental change, masking clouds and finding relevant satellite scenes across large archives. The conversation explains how embeddings turn satellite imagery into representations that AI models can use to identify patterns. The team also discusses the differences between patch-based and pixel-based models, helping you understand how each approach supports different analytical tasks. Looking ahead, the speakers explore what multimodal models, AI agents and onboard satellite processing could bring to geospatial AI (GeoAI). They also address the challenges of working with large datasets and the role of open data and shared frameworks in making these approaches easier to adopt. Whether you’re developing geospatial applications, monitoring environmental change or exploring AI for satellite data, watch the episode for practical insights to inform your next project. What you’ll learn: • How embeddings help turn satellite imagery into a resource for AI analysis • When to use patch-based and pixel-based models • How anomaly detection can help identify wildfires without labelled examples • Why cloud masking remains a challenge for EO workflows • How text-to-image search helps you find relevant satellite scenes • What emerging GeoAI approaches could mean for your future workflows Speakers: Marcin Kluczek - EO Algorithms Lead at CloudFerro Bartosz Augustyn - Junior Data Scientist at CloudFerro Bartłomiej Ostrowski - Junior Data Scientist at CloudFerro