SAFAARI Framework Boosts Advertiser Response Intelligence with Schema-Aware AI.

Customer support systems are increasingly using agentic chatbots, but these often struggle to access complex enterprise data without pre-defined API endpoints. The SAFAARI framework addresses this by employing specialized agents for content, metadata, and orchestration to streamline the critical process of schema linking in Natural Language to SQL (NL-to-SQL) systems. This innovation allows chatbots to more effectively query and utilize diverse data sources. The framework also introduces SEAL, a novel composite metric for evaluating system performance, which penalizes inconsistent results. Through rigorous testing, SAFAARI achieved an 81.66% SEAL score, representing a 6.65% improvement over baselines, with notable gains in data accuracy and schema-linking precision. This efficiency translates to an 8x reduction in development time for API creation and enhanced self-service capabilities, particularly beneficial for organizations with intricate data ecosystems.