Validating UGI’s transition to LLMs
Co-one tested chatbot intents across user profiles and grouped response errors by model version as Garanti BBVA moved UGI to large language models.

Co-one Insights
Customer stories, technical perspectives, and the lessons behind dependable AI systems.
Selected enterprise stories
Explore how Co-one has helped enterprises evaluate AI systems, improve intelligent applications, and solve complex data challenges.
Co-one tested chatbot intents across user profiles and grouped response errors by model version as Garanti BBVA moved UGI to large language models.


Co-one enriched the catalog with consistent fabric and pattern tags so shoppers could find items by the attributes they cared about.
Co-one labeled customer queries and cross-checked annotations to help the chatbot handle ambiguous requests.
Technical perspectives

Featured perspective
Logs and traces show what happened. Evaluations help establish whether an agent’s actions met the task and its requirements.
Read by topic