AI Lesson Planning Done Right: A Complete Real Example

Most AI lesson plans are thin. Ask a general chatbot to plan a lesson and you get a tidy list: objectives, a warm-up, maybe a worksheet if you push. It's the kind of thing a teacher writes in fifteen minutes and would never actually hand to a colleague.

That gap is what I've been trying to close. I wanted an assistant that produces the material an experienced teacher actually makes — the whole unit thought through, with assessment that shows what students understand rather than what they memorized.

This post shows what that looks like. No mock-up. The example below is real output from Methodics AI, generated for Grade 8 mathematics, Common Core, topic Linear Functions.

What usually goes wrong with AI lesson plans

Ask for a lesson and you get a list, not a plan. "Objectives: students will understand fractions. Warm-up: review. Activity: worksheet. Assessment: quiz." It doesn't tell you what to say at minute 35, what misconception almost every student walks in with, or how to handle the kid who finishes in ten minutes.

A real unit has to answer harder questions. What learning outcome does this map to in the standard? Where does the class usually stall? Where does AI help the student, and where does it quietly hurt? What does the teacher do while the students are working?

What a complete unit contains

Methodics AI builds a unit in six sections:

The deliberate choice: it develops one lesson fully instead of ten lessons shallowly. A lesson you can teach tomorrow beats a framework you still have to finish yourself.

The real example: Linear Functions, Grade 8

Config: English, Common Core, Mathematics, Grade 8, two 60-minute lessons, USA context, topic Linear Functions. Here is what came back. The example below is an excerpt; download the complete sample for the full 54,000-character unit, all six sections, ready to read or print.

Overview (from the generated unit):

This 2-lesson unit builds conceptual understanding, procedural fluency, and real-world transfer of linear functions in multiple representations (equations, tables, graphs, verbal models) through active inquiry, physical experiments, and structured differentiation.

2 lessons × 60 min · 28 tasks with answers · 3 differentiation profiles · 10 typical errors · 1 assessment system

The full lesson, minute by minute (this is the "one lesson done to the end"):

Lesson flow — 60 minutes

0–10 min · The Hook: "The Pizza Delivery Dilemma"

You want to order 4 large pizzas. Option A charges a $2 per-pizza delivery fee plus $12 per pizza. Option B charges a flat $10 delivery fee plus $10 per pizza.

The teacher asks, without a formula yet: which is cheaper for 1 pizza? For 4? For 8? And why did the cheaper option switch as the order grew?

Why this matters: it creates the cognitive conflict that justifies linear models. A one-time fee can overpower a lower per-unit rate for small quantities — that's the whole point of a y-intercept.

10–25 min · Discovery: The Leaking Water Bottle (hands-on)

A clear 1-liter bottle, marked in centimeters and filled to 24 cm, with a small hole near the bottom. Start the timer and call out the water height every 10 seconds.

Students write the equation h = −0.4t + 24, then solve 0 = −0.4t + 24 → t = 60 seconds. The phenomenon comes first; the formula comes after. No special lab needed — just a bottle, a tack, and a tray.

25–45 min · Guided practice: The Ride-Share Pricing Battle

Two Chicago ride-share companies. UrbanCar gives the equation y = 1.50x + 5.00 directly; MetroCab gives a table of costs instead. Pairs have to recover MetroCab's equation (y = 1.50x + 8.00), graph both, and answer the real question: since both have the same slope, will MetroCab ever be cheaper?

The answer lives in the y-intercept, not the slope. Equal slopes mean the lines never meet, so the one with the lower starting fee is always cheaper.

45–55 min · Error analysis: Spot the Impostor

A fictional student "Alex" solves: "a gym charges a $40 start-up fee and $20 per month; find the cost for 6 months." Alex writes y = 40x + 20 and gets $260. The class spots the flaw — Alex swapped the slope and the initial value. Correct: y = 20x + 40, so $160. Alex overcharged by $100.

55–60 min · Exit ticket

A phone plan charges $30 per month plus $0.05 per text. Write the cost equation and explain what the slope means. (Answer: C = 0.05n + 30; each text adds 5 cents.)

The unit even ends the lesson with a "why this lesson is excellent, not average" card. Its four reasons: the phenomenon precedes the formula, the recurring rate is contrasted with the one-time fee, parallel lines are discovered rather than asserted, and "Alex's" mistake turns a misconception into a teaching moment.

Assessment: The Food Truck Startup

A taco truck owner in Austin must choose between two kitchens. Kitchen South charges a $400 monthly deposit plus $15 an hour. Kitchen North has no deposit but charges $35 an hour.

Students (a) write both linear functions, (b) compute the cost at 12 hours, (c) invert the problem — a $1,000 budget means how many hours?, (d) find the break-even point, and (e) decide which kitchen to rent for roughly 24 hours a month. Every sub-question ships with a full answer, and a scoring table spells out exactly what full credit, partial credit, and no credit look like.

And the worksheet is a printable three-tier handout — level 1 "basic foundations, mandatory for all," level 2 intermediate, level 3 extended challenge — with Name/Date/Class fields and room to work.

What makes this different from a chatbot

The prompts encode a teaching methodology, not just "write a lesson plan." Two things stand out.

First, assessment that reveals thinking. The material uses a 60/40 principle: most of the grade comes from showing reasoning, not repeating facts. Tasks ask students to interpret, invert, and interpret again — the way an exam actually checks understanding rather than recognition.

Second, honest AI boundaries. The unit tells you where AI helps (generating practice problems) and where it harms (a student getting answers without the reasoning), plus the rule that real student names and data never go into the tool.

The honest limitations

AI-generated material is a starting point, not a finished product. Every unit ends with "review and adapt to your students." The methodology also refuses to invent standard codes — if a curriculum reference isn't certain, it's stated in words instead of a fake code. That's on purpose. Fake specificity in front of a class is worse than honest generality.

How to try it

1. Buy access at Methodics AI — €19.99 one-time for 15 complete units.
2. Open the assistant, enter your key, and set language, standard, subject, and grade.
3. Generate, download as HTML, print to PDF.

Or start free: the assistant gives you one complete unit per day, no key needed.

FAQ

Is the output really this long?
Yes. The full unit behind this example runs about 54,000 characters across all six sections. The lesson above is an excerpt; the rest is assessment, differentiation, the teacher's handbook, and the worksheet. Download the complete sample to see it all.

Does it work with my country's curriculum?
It adapts to any standard you type in — Common Core, IB, Cambridge, UK National Curriculum, Skola2030, and more. Universal mode fits any standard; European mode adds EU key competences, DigComp, and GDPR.

Can students misuse it?
The material itself is the safeguard. Each unit includes a section on where AI harms learning and a data-privacy note, so the boundaries are embedded in what you hand out.

What if I don't like a generated unit?
Regenerate with different instructions. Because you review before you use, you keep the final say.

Try the assistant → — one complete unit is free, every day. Get your access key at miskola.lv/methodics-ai.html.

Methodics AI by Miskola — Gatis Šeršņevs. AI-generated content; always review and adapt to your students.