Debug by symptom
Your AI feature is misbehaving. Which of everything you've learned actually fixes it?
The idea inside
Each failure symptom points to one mechanism, and the fix you already know.
After this lesson
You can map an LLM failure symptom to its mechanism and the right fix (retrieval, tools, trimming context, thinking budget, constrained decoding).
Where it leads
Inside this lesson
That's the real lesson stage, paused. Claim your pass to operate it.
See how AI actually works, end to end.
This lesson is one stop on the full arc. Unlock all of it, and keep it for life.
What you get
- The 34-lesson main path, a finishable route from a word to agents
- Goal tracks for using AI at work and building AI features
- Boss labs that make you apply a whole act, not just recognize it
- Spaced recall that brings each idea back before you forget
- Course memory: every term defined, with links to where it first appears
- A shareable capability card when you finish the main path
- Lifetime access on every device, every future lesson included
Not videos to watch. You predict, operate the machine, then prove it. That is why it stays.
99 interactive lessons and challenges. No videos, no code.
Free launch pass: lifetime access, no card needed
New here? The first lessons are free to try. Start with lesson 0.1
What this lesson shows
Each failure symptom points to one mechanism, and the fix you already know.
The question it opens with
Your AI feature is misbehaving. Which of everything you've learned actually fixes it?
The walkthrough, in the lesson's own words
- Repaired. Roll the next broken assistant onto the bench.
- A broken assistant is on the bench. Place the tool that repairs it.
- Work through the bench: repair at least 3 of the 5 broken assistants.
- Every failure traces to a mechanism and a fix you already learned.
- You don't debug an LLM by poking at it. You read the symptom, name the mechanism behind it, and reach for the fix that targets that mechanism.
- Every fix here is something you already met: retrieval, tools, trimming context, the thinking budget, structured outputs. Debugging is just routing a symptom to the right one.
- The wrong fixes weren't random, each is a real tool that simply targets a different mechanism. More retrieval can't stop overthinking; a tool can't ground a missing fact; trimming can't reshape prose into JSON. Naming the mechanism is what tells you which lever to pull, so you stop guessing and start diagnosing.
- Your assistant confidently cites a refund policy that doesn't exist. A teammate wants to lower the temperature. Will that fix it?
- No: a made-up fact is missing knowledge, not drift, so temperature is the wrong lever. Read the symptom to its mechanism, here a plausible token with no source, and reach for retrieval (RAG) so it quotes the real policy instead of inventing one.
Key takeaway
You can read a symptom and reach for the right fix instead of guessing.
What you can do after this lesson
You can map an LLM failure symptom to its mechanism and the right fix (retrieval, tools, trimming context, thinking budget, constrained decoding).
This is the written summary. The lesson itself is interactive: you predict, drag and operate the mechanism above, and the reveal answers you.