Applied AI

Course map · S1 How LLMs work · Episode 1 of 12

One token at a time

S1 E1 Explainer · 13 slides · 28 September 2026

  1. Slide 1 of 13, Cover.
  2. Slide 2 of 13, Why it matters.
  3. Slide 3 of 13, Mental model.
  4. Slide 4 of 13, Concepts.
  5. Slide 5 of 13, Concepts.
  6. Slide 6 of 13, Concepts.
  7. Slide 7 of 13, Where it breaks.
  8. Slide 8 of 13, For engineers.
  9. Slide 9 of 13, For engineers.
  10. Slide 10 of 13, In practice.
  11. Slide 11 of 13, For leaders.
  12. Slide 12 of 13, Decision rule.
  13. Slide 13 of 13, Series map.

Lesson notes

What a large language model (LLM) actually does, and why every call starts fresh.

Decision rule

Treat every model call as stateless and probabilistic: send the context it needs, and verify what it returns.

Ideas in this lesson

  • Tokens
  • Next-token prediction
  • Statelessness
  • Probability
  • Context

Take it further