Where this course goes (and where you fit) — step 4 of 9
This is not a 10-day bootcamp
The most common lie in AI education in 2026 is the timeline. "Learn AI in a weekend." "Build your first agent in 48 hours." "From zero to AI engineer in two weeks." None of that is honest. Real competence takes three to six months of consistent work, and the only way to make that shorter is to start from a higher floor.
Here is what actually happens, month by month, for someone starting at ch01 with no programming background and 10-12 hours a week to give this.
Month 1 — ch01-04
You finish variables, functions, lists/dicts, and loops. That's the language fluency floor. You can type a 10-line program. You feel slow and awkward and the syntax trips you up constantly. The error messages still feel rude. You make six mistakes per session and only see why on the fourth one.
This is the hardest month. It is harder than ch26. Push through.
Month 2 — ch05-11
Conditionals, tracebacks, mutation, modules, error handling, files,
classes. Your programs grow from 10 lines to 30 to 60. You start
seeing patterns repeat — for x in y, if x is None, try/except.
Tracebacks stop feeling rude and start feeling helpful, which is the
moment Python actually flips for most people.
By the end of month 2 you have written about 400 lines of code in total. None of it does anything impressive. That's fine — that's not what this part is for.
Month 3 — ch12-19
Your first LLM API call (ch13). Your first prompt that runs as code (ch19). Somewhere in here, usually around ch16 (agent loops), is the "I built something" moment — when a Python script you wrote calls a model, gets a response, decides what to do, and calls the model again. That moment matters more than any chapter completion certificate.
By the end of this chapter you will have graded an LLM's work and been right about something it got wrong. That is the job now.
This is the phase where you start to feel like a builder. It is also the phase where most people start telling their friends, "I'm learning to build with AI," and the friends say "oh cool" and have no idea what's happening. Do it anyway.
Month 4-5 — ch20-25
Agent traces, evals, retrieval, production tradeoffs, debugging, and the capstone. This is the chapter sequence where the demo dies and the product begins. You will write something that breaks in production. You will spend a Saturday figuring out why. You will fix it. That Saturday is the day you become an engineer.
By the end of the capstone you have shipped something a friend could use. Not great, not polished. Real.
Month 5-6 — ch26-30
The harness-engineering ramp. Agent harnesses, image gen, video gen, programmatic design, and then ch30, which is the chapter that names out loud what most of the other chapters were quietly teaching: how to design the scaffolding around a model so it stops being a demo and starts being a system.
Most people don't get here. The people who do are the ones who become senior at this from a standing start.
Two honest calibrations
If you came in already coding — strip a month or two off the front. You will sprint through ch01-11 in a couple of weeks. Your bottleneck is not syntax; it's the prompting / evals / harness side. Aim for ch12-25 in 6-8 weeks and ch26-30 in another 6-8.
If you have very limited time — 5 hours a week or less — the timeline doubles. That is honest. The skill still compounds, but you'll hit the "I built something" moment around month 6 instead of month 3. That is not failure. It is a 5-hour-a-week schedule. Plan your life around it instead of pretending you'll suddenly have more time next week.
The course is not "AI bootcamp in 10 days." It is three to six months of real work, depending on starting point. Anyone selling you faster is selling you something else.
Where this course goes (and where you fit) — step 4 of 9
This is not a 10-day bootcamp
The most common lie in AI education in 2026 is the timeline. "Learn AI in a weekend." "Build your first agent in 48 hours." "From zero to AI engineer in two weeks." None of that is honest. Real competence takes three to six months of consistent work, and the only way to make that shorter is to start from a higher floor.
Here is what actually happens, month by month, for someone starting at ch01 with no programming background and 10-12 hours a week to give this.
Month 1 — ch01-04
You finish variables, functions, lists/dicts, and loops. That's the language fluency floor. You can type a 10-line program. You feel slow and awkward and the syntax trips you up constantly. The error messages still feel rude. You make six mistakes per session and only see why on the fourth one.
This is the hardest month. It is harder than ch26. Push through.
Month 2 — ch05-11
Conditionals, tracebacks, mutation, modules, error handling, files,
classes. Your programs grow from 10 lines to 30 to 60. You start
seeing patterns repeat — for x in y, if x is None, try/except.
Tracebacks stop feeling rude and start feeling helpful, which is the
moment Python actually flips for most people.
By the end of month 2 you have written about 400 lines of code in total. None of it does anything impressive. That's fine — that's not what this part is for.
Month 3 — ch12-19
Your first LLM API call (ch13). Your first prompt that runs as code (ch19). Somewhere in here, usually around ch16 (agent loops), is the "I built something" moment — when a Python script you wrote calls a model, gets a response, decides what to do, and calls the model again. That moment matters more than any chapter completion certificate.
By the end of this chapter you will have graded an LLM's work and been right about something it got wrong. That is the job now.
This is the phase where you start to feel like a builder. It is also the phase where most people start telling their friends, "I'm learning to build with AI," and the friends say "oh cool" and have no idea what's happening. Do it anyway.
Month 4-5 — ch20-25
Agent traces, evals, retrieval, production tradeoffs, debugging, and the capstone. This is the chapter sequence where the demo dies and the product begins. You will write something that breaks in production. You will spend a Saturday figuring out why. You will fix it. That Saturday is the day you become an engineer.
By the end of the capstone you have shipped something a friend could use. Not great, not polished. Real.
Month 5-6 — ch26-30
The harness-engineering ramp. Agent harnesses, image gen, video gen, programmatic design, and then ch30, which is the chapter that names out loud what most of the other chapters were quietly teaching: how to design the scaffolding around a model so it stops being a demo and starts being a system.
Most people don't get here. The people who do are the ones who become senior at this from a standing start.
Two honest calibrations
If you came in already coding — strip a month or two off the front. You will sprint through ch01-11 in a couple of weeks. Your bottleneck is not syntax; it's the prompting / evals / harness side. Aim for ch12-25 in 6-8 weeks and ch26-30 in another 6-8.
If you have very limited time — 5 hours a week or less — the timeline doubles. That is honest. The skill still compounds, but you'll hit the "I built something" moment around month 6 instead of month 3. That is not failure. It is a 5-hour-a-week schedule. Plan your life around it instead of pretending you'll suddenly have more time next week.
The course is not "AI bootcamp in 10 days." It is three to six months of real work, depending on starting point. Anyone selling you faster is selling you something else.