Kong to Support AGC in Strengthening Multi-LLM Governance and Cost Optimization

Kong Inc. announced that AGC Inc. has adopted its API Management and AI Gateway services as part of a company-wide initiative to support secure AI utilization and improve development efficiency. This adoption is part of an effort to establish a common AI development foundation for the entire company, with an eye toward future expansion to overseas locations.

Background: Acceleration of Multi-LLM Utilization and Associated Challenges As interest in AI grew within AGC, several challenges were identified. There was a rising demand from development teams to flexibly use various LLMs, such as Gemini and Anthropic, in addition to the mainly used Azure OpenAI. However, individual contracts for each business unit complicated processes and hindered rapid, company-wide AI adoption. Furthermore, it was difficult to accurately track API usage—who was using what and how much—making overall cost management and governance challenging. Anxiety about 'high bills from overuse' also acted as a psychological barrier for users, discouraging active AI experimentation.

Key Reasons for Selecting Kong: Balancing Governance and Speed AGC selected Kong to address these challenges for the following main reasons:

- Centralized Management of Multi-LLM (AI Gateway): Kong acts as an intermediary, standardizing different LLM APIs, allowing developers to choose and switch between optimal AI models through a single interface with minimal switching costs. - Consolidation of Contract and Approval Processes: By using Kong as a pre-vetted common platform, contract and governance management are centralized, streamlining processes for individual business units. - Detailed Visualization of Usage and Costs: The platform enables tracking of usage per API consumer (department, project, or individual), facilitating future automation of internal chargebacks. - Guardrails to Prevent Budget Overruns (AI Rate Limiting Advanced): The ability to set quotas and rate limits for each consumer automatically controls usage to prevent budget excesses, alleviating user concerns about unexpected high costs. - Advanced Support for Agents/MCP: The platform's high extensibility supports AGC's vision of developing AI agents that access internal data and research papers.

The solution was procured through AWS Marketplace for its flexibility and alignment with existing cloud procurement processes.

Future Outlook The project will begin with a rollout to developers and will be expanded in phases.