A new way of learning

Tutorials & labs open soon

Animated tutorials with labs you run yourself.

readcutaway.com

  • Watching endless tutorials
  • Reading endless blogs
  • Memorizing definitions
  • Building a demo that works once
See what you'll learn to decide
Why Cutaway Real engineering decisions

AI engineering is making the call.

The questions that come up after the demo, when there are real users, a budget and an on-call rota. Every volume answers four of them: why, how would I design it, how would I debug it, and how would I take it to production.

Before you build
Should this be AI at all?
The trapAdding a model because you can, when a rule or a search box would do.
Answered in Should This Be AI?
Choosing models
Which model, and what happens when it changes?
The trapTrusting the provider's benchmark instead of your own test set.
Answered in Choosing and Changing Models
Knowledge
Why RAG, and what if the retrieved context is wrong?
The trapBlaming the model for an answer the retrieval never gave it.
Answered in Retrieval Engineering
Agents
Should it be an agent at all?
The trapAdding another agent to fix a reliability problem.
Answered in Agent Architecture
Evaluation
How do you know it's working?
The trapAn eval that reports everything is fine, because it can't see the hard cases.
Answered in Eval Engineering
Cost
What does one user cost, and what can you cut?
The trapCutting the tokens that were carrying the answer.
Answered in Token Cost Engineering
Security
What happens when someone attacks it?
The trapOne guardrail at the door, checking evidence that doesn't exist yet.
Answered in Agents Guardrail
Production
How will you debug it when real users arrive?
The trapLogging the answer, but not what was retrieved or which version produced it.
Answered in Shipping and Debugging
Find your track →
Start here Choose your track

Pick where you are.

Guided paths, each read in order, from your first idea to the decisions your job asks of you. Pick the one that fits.

See it · Concepts · free

Every idea, animated.

One page per idea, from tokens and the KV cache to prompt injection. Watch each mechanism move, at your own pace.

All concepts →
Run it · Tutorials

Go deep. Then build it.

Animated, step by step, from how the mechanism works to when to stop. Every tutorial comes with labs you run on your own machine.

Labs plain Python · offline

Then run it yourself.

Every tutorial ships with labs: small, plain-Python programs that reproduce the mechanism on your machine, in under a second, with no API key. This is real output from the Eval Engineering lab.

eval-engineering/labs/answer

  
  
Real output from the Eval Engineering labs. Precision 1.00, recall 0.59: every answer it flags as wrong really is wrong, but it misses 41% of the wrong answers. Plain Python, runs offline in under a second.
Decide · For engineering leaders

The questions you'll be asked.

Your team ships with agents now. Short, standalone volumes, every figure marked as cited, modeled or an exhibit, so you can repeat it in a meeting.

Be first in when it opens.