Okta has introduced an identity-scoped Model Context Protocol tool that can help reduce costs associated with using AI agents. The company's new technology, known as SCPing (Scope for Computing Purpose), allows models to query multiple tools and services while only consuming a limited amount of computational resources.
Each time an AI agent makes a call to the model, it includes schema information for every exposed tool, along with names, descriptions, and parameters. This results in higher-than-expected computational overhead, known as the "tool tax," which can significantly increase the cost of using these AI agents. According to Okta, SCPing reduces this tool tax by limiting the number of tools queried at a time.
By implementing SCPing, businesses can reduce their costs associated with using AI agents and scale more efficiently. The company's technology is designed to be flexible and adaptable to different use cases and deployment environments. With SCPing, model developers and data scientists can focus on developing their AI solutions while keeping costs under control.