AI API fundamentals, understand how to talk to Claude under the hood

You use Claude via a UI (Claude Code, Workbench, Lovable…) but want to understand what happens when an agent calls the model? This course lays the foundations: making a first API call, choosing between Haiku/Sonnet/Opus with dated snapshots, tuning parameters (max_tokens, temperature, stop, system prompt), streaming the response for UX, and going multimodal with vision. By the end, you read a Claude API call autonomously and know how to arbitrate each setting.

Level : beginner

Modules

  1. 1. Making your first Claude call, SDK + messages format
  2. 2. Picking the right Claude model, dated snapshots vs moving aliases
  3. 3. Tuning parameters, max_tokens, temperature, stop, system
  4. 4. Streaming, prod UX and Time to First Token
  5. 5. Vision, going multimodal with image + text

View on Coeurdar