Story 2. The Past: Building the Foundation 3. The Present: Innovation Culture, Automation and the Agentic AI Platform 4. Our Technical Challenges and Strategies 5. The Future: AI Hub 6. Q&A
and technology 10 The Past: Building the Foundation • Timing with ChatGPT was ideal • Business trying out ChatGPT to solve their own issues • Very quick to get to 70% value, but • How to make it production worthy? • Requires process, evaluation and integration • Innovation to get to the next maturity phase
platform AI enablement at TBA Data and AI teams – Maximum Control • Full flexibility and customization • Fine-tune, integrate, and govern AI Business Colleagues – Low barrier • Easy-to-use interfaces • Safe experimentation without tech deep-dives • Empower everyday AI use
1 2 3 4 Controlling AI in production Code reusability across agents Multi-language and multi- currency predictions Error vs user adoption Strategies to mitigate Challenges Agent serving as an API PyFunc conda environment and python wheels Relentless experimentation and AB testing Agent state management and multilanguage LLM embeddings
Future: AI Hub AI HUB Predicted Lot Value 01 05 Similar Lots 04 Lot Descriptor 02 Lot Categorizer 03 Negotiations Agent • Inspiration from ChatGPT to scaled AI • Quick iteration with engaged business colleagues • Use a quick R&D mindset to validate value rapidly.