New Delhi, June 2 -- A weaker rupee has made dollar-priced artificial intelligence (AI) services such as ChatGPT and Claude 10-15% more expensive for Indian startups, pushing founders to adopt both technical and business strategies to control costs.

One way to reduce expenses is by optimizing context memory, a feature that allows AI systems to retain previously generated information, user preferences, and frequently used responses. Instead of sending every query or prompt back to a large language model (LLM), the system can retrieve relevant answers from stored context, reducing token consumption and lowering API costs.

"For example, if a question has already been answered, the system can pull the response from context memory instead of...