Applied AI

A free course by Brendan Kelly

Learn how AI actually works.

Short visual lessons, from the first token to agents in production. Every claim is tested in code you can run, and every lesson ends with one rule you can use at work the next day.

Start the course Browse S2: Retrieval and RAG
  • 16 lessons live
  • 15 seasons mapped
  • All code published
Lesson Coming soon Idea Season Hover to explore, drag to rearrange

Inside every lesson

A handful of slides. One idea that sticks.

  1. Example slide from S2 E1, Why retrieval exists: a mental model you can picture.

    A mental model you can picture

    Each idea starts as one clear picture, not a wall of jargon.

  2. Example slide from S2 E1, Why retrieval exists: tested with real data.

    Tested with real data

    Claims are measured, with the code published so you can rerun them.

  3. Example slide from S2 E1, Why retrieval exists: one rule to keep.

    One rule to keep

    Every lesson ends with a decision you can apply the next day.

Slides from S2 E1, Why retrieval exists.

What you'll take away

Rules worth keeping

  1. S1 E10

    Turn reasoning on for problems that need several steps held in mind at once. Leave it off for lookups and simple sums, and pilot before you commit.

  2. S1 E11

    Keep reasoning at its lowest setting by default. Turn it up per task, only where a paired test on your own items shows it earns its tokens and seconds.

  3. S1 E12

    Send the model what it needs, verify what it returns, and trust a measured rate over any single answer.

  4. S2 E0

    Build the simplest complete system. Measure it. Change one stage at a time, and keep only what clears a bar set in advance.

  5. S2 E1

    Use retrieval when the answer lives in your own data. Then log what it retrieved, and measure search and answer separately.

  6. S2 E2

    Build the simplest complete pipeline and measure it before tuning any stage.

Every lesson ends with one. See them all, lesson by lesson

Where the course goes

From how models work to how to lead with them

Two technical seasons, then a leadership season that applies them.

  1. S1 How LLMs work Technical
  2. S2 Retrieval and RAG Now running
  3. S3 AI strategy and use case selection Leadership
  4. S4 Adapting models Technical
  5. S5 Agent fundamentals Technical
  6. S6 AI economics Leadership
  7. S7 Data for AI Technical
  8. S8 Evaluation Technical
  9. S9 Building AI teams and operating models Leadership
  10. S10 Agents in production: reliability and traceability Technical
  11. S11 Agents in production: security Technical
  12. S12 Delivery, adoption, and change Leadership
  13. S13 Agents in production: performance and cost Technical
  14. S14 Agents in production: resilience Technical
  15. S15 Governance and the regulated EU enterprise Leadership

Who's teaching

Brendan Kelly

Brendan Kelly is a Director of Software Engineering with more than 20 years across engineering and engineering leadership. Applied AI is where he learns in public. Every episode tests a claim against published code and results, and ends with a decision rule worth keeping. When a claim fails, the episode says so.

See the code behind every lesson

Start at the beginning. Each lesson takes a few minutes.

Start the course