The Korea Agro-Fisheries & Food Trade Corporation (aT) is upgrading its ChaTi conversational information search service within the KATI platform by adding agentic AI to help exporters navigate country-specific regulations, labeling, and non-tariff barriers.

Bespin Global has won the contract to upgrade ChaTi, expanding its capabilities for agricultural and food export information under the KATI platform.

The upgrade prioritizes security and reliability, incorporating data masking, prompt injection defenses, and sources verification by showing original reports and URLs, in line with standards like the Digital Service Professional Contracting System and GS certification.

The project will deploy multi-agent orchestration, enabling specialized agents in market analysis, regulatory assessment, and data analysis to collaborate on complex queries.

CEO Heo Yang-ho says the project demonstrates Bespin Global’s AI technology applicability to public-interest domains, leveraging Public AX experience to build a trusted AI platform for government institutions.

The enhanced ChaTi will let exporters and distributors interactively review country-specific market trends, statistics, and regulatory requirements, reducing search time and aiding market and product decisions.

Key features include access to export statistics databases and non-tariff barrier data, OCR-based data extraction from PDFs, tables, and images, and coordinated multi-agent collaboration to answer export-related questions.

Bespin Global emphasizes its public-sector track record and positions this upgrade as expanding into the agricultural and fisheries food distribution sector, building on deployments in environment, safety, tourism, and education.

The upgrade uses Bespin Global’s HelpNow AI platform to build a next-generation intelligent assistant that analyzes user needs and executes tasks to help exporters access information and make faster, better-informed decisions.

Bespin Global has been selected to implement ChaTi’s enhancement, applying HelpNow AI to transform ChaTi into an agentic AI-based assistant.

The initiative supports expanding K-food exports, addressing growing information needs like country-specific quarantine and labeling rules, with last year’s K-food exports at $13.62 billion and a government target of $16 billion for 2026.

Technical upgrades include enhancing ChaTi’s LLM and retrieval-augmented generation architecture, expanding training data with internal databases (export statistics, non-tariff barriers), and implementing OCR to extract data from unstructured sources.