An API (Application Programming Interface) is how software communicates with an AI model. When a company integrates AI into their product, they don't install the model on their server — they send requests to the AI provider's API and get responses back. For example, a customer service app might send a customer's question to the Claude API, receive the AI-generated response, and display it in the chat window. API pricing is typically per-token (the amount of text processed). Understanding API costs is essential for budgeting AI implementations — a high-volume application can cost significantly more than the per-user subscription price suggests.