Category: API
-
RAG Cost Optimization
The article discusses how costs associated with Retrieval-Augmented Generation (RAG) systems can spiral due to fragmented control and local decision-making in AI workflows. It emphasizes the need for a centralized control layer to manage resource allocation, caching, retrieval depth, and model calls, ultimately aiming to reduce unpredictability in AI spending.
-
When Prompt Injection Gets Real: Use GraphQL Federation to Contain It
In 2024-2025, AI security incidents revealed that existing controls designed for human users are inadequate for large language models (LLMs). Vulnerabilities stemmed from unverified model execution, leading to data leaks and compromised systems. WunderGraph Cosmo proposes a federated architecture to establish runtime boundaries, ensuring safer execution and access governance, ultimately enhancing AI security.
-
Over 15 Years of APIs with Kevin Swiber
Kevin Swiber discusses API governance, AI’s role in developer tools, and the Model Context Protocol (MCP) in a recent episode of The Good Thing. Emphasizing the importance of stable infrastructure, he highlights how autonomous teams can lead to sprawl and the continuous need for APIs. Ultimately, he advocates for boring yet effective infrastructure.