Course Arc
A semester course in directing AI on purpose.
18 weeks · 75 minutes a day, 5 days a week · one credit
This course is about doing real work with AI tools - not consuming what they make, but directing what they make. The skill is called harnessing: putting AI to work on the problems you actually have, in language clear enough that the machine can act on it. By the end of the semester, you’ll have a personal site on the internet with a working tool you built in response to what your school community said they need.
You’ll work on your site every week. Some days you’ll learn something new and apply it. Some days you’ll iterate on what’s already there. Some days you’ll fix something that broke. All three are real work, and you’ll do all three across the semester.
A harness has two ends. At the prompt end, you direct AI through clear specifications. At the system end, the tools you use - and the ones you build - are themselves harnesses on what AI can do. This course meets you at the prompt end and opens the door to the bigger systems-level ideas.
Most people who use a computer use it for a pretty narrow band of things: email in Gmail, papers in Google Docs, and then a lot of consumption - TikTok, Instagram, Roblox, YouTube. That's the front end of technology: the polished surface someone else built for you to use.
This act is designed to take you from that front end - the user-facing side of the screen - to the back end, the world where developers actually do the work of building what everyone else uses.
What you’ll make
A personal landing page, on the internet, at a URL you can share. Anyone with the link can visit your page. It has your name on it, your voice in it, and reflects what you care about.
The arc of Act 1
There's a real computer in this classroom running a real AI model. This week is about understanding it. You'll learn what's inside a computer (CPU, RAM, SSD, motherboard, GPU, VRAM) and why the specific hardware matters for what AI can do. You'll learn what AI actually is: neural networks, training, freezing, inference, fine-tuning - not in deep math, but enough to know what's happening when you use these tools. You'll learn the difference between AI in the cloud (like ChatGPT) and AI on your own machine (like the one in this room), and why that difference matters for your data and your work. You'll learn about speculative decoding - the technique that makes the model in this room produce code as fast as it does. By the end of the week, you'll have interacted directly with the model through its raw interface, before you ever see the polished tool.
You'll learn what a prompt is and what makes a prompt effective. You'll learn the difference between vibecoding (winging it with vague instructions) and spec-driven development (writing precise specifications that produce predictable output). Mid-week, you'll get an account on the muggsofcode build tool and start using it. The tool has three viewers: a markdown spec editor, a code viewer where the AI's output appears, and a page viewer where the rendered result shows. You'll learn to write specs in the tool's four-section format (Overview, Structure, Style, Behavior). Your website observation practice continues this week.
You'll spend this week learning to read what the tool produces. HTML, CSS, what each does. You'll iterate on specs in the tool and observe how changes to your spec ripple through to the code and the page. You'll meet GitHub conceptually - what version control is, what a repository is, why working developers use it. Your website observation practice continues.
This is when your real landing page starts coming together. You'll spend most of class time working in the tool on your own page. The deployment path gets introduced - how a page in the tool becomes a page on the internet. First deployments happen this week.
The Act 1 commitment is firm: every student leaves Act 1 with a deployed landing page. This week is about making sure that happens. Students who need help getting their page live get it. Students who are already deployed iterate and refine. By the end of Week 6, every student has a URL on the public internet with their landing page at it. Act 1 ends with a brief reflection and a preview of what Act 2 will add to your site.
What Act 1 teaches
You'll learn how to articulate intent in writing clearly enough that an AI can act on it. You'll learn how to read code and how to recognize when an AI has produced what you asked for versus something different. You'll learn the basic infrastructure of how a webpage gets from your screen to the internet. You'll learn what the machine running these tools actually is, and why that matters. You'll have an ongoing writing practice through the website observation work. And you'll start developing a relationship with AI that's based on direction rather than consumption.
What you’ll have at the end of Act 1
- ◆A landing page deployed at a URL you can share
- ◆An understanding of what's inside the machines that run AI
- ◆A class AI policy you helped write
- ◆Five weeks of website observations that have built your eye for design choices
- ◆The beginning of a working relationship with the muggsofcode tool
What you’ll make
Your landing page grows into a three-page site. Navigation gets added so visitors can move between pages. Two new pages get built: an Inspiration & Rationale page (where you analyze the school survey data and articulate what you'll build in Act 3) and a Tool page (where the Act 3 tool will eventually live).
The arc of Act 2
This is the data act. Across these six weeks, you'll learn what data actually is, how it comes in different formats (markdown, txt, json, csv, spreadsheets, a touch of SQL), and how to work with each. You'll learn about surveys - what they're for, how to design them, what kinds of data different question types produce. The class will collaboratively design a survey for the school community, asking what tools they wish existed to help them with their work. You'll administer the survey, collect responses, learn to clean and analyze the data, and produce visualizations of what you found.
The output of this work goes on your Inspiration & Rationale page. That page tells the story: here's what the school said it needs, here's what I care about, and here's where those overlap. The page also articulates what you're going to build in Act 3 and why.
Alongside all of this, you'll add navigation to your site and stand up the Tool page (mostly placeholder during Act 2 - it gets filled in Act 3).
Act 2 will get its own detailed page when the week-by-week plan is ready.
What Act 2 teaches
How to design surveys and recognize what each question type produces. How to work with data in multiple formats. How to analyze results and communicate findings through visualizations. How to build a multi-page site with consistent navigation. How to use evidence - what your community actually said - to justify what you'll build.
What you’ll have at the end of Act 2
- ◆A three-page site with working navigation
- ◆A completed school survey with results
- ◆Data visualizations on your Inspiration & Rationale page
- ◆A clear rationale for the tool you'll build in Act 3
- ◆A standing-up Tool page ready to receive the tool
What you’ll make
A working web tool, built in response to what the school survey revealed, deployed at the Tool page you stood up in Act 2.
The arc of Act 3
This is the building act. You'll scope your tool based on what Act 2's data showed and what fits your interests. You'll spend weeks designing and building it, using the muggsofcode tool and whatever additional code work the tool requires. You'll style it, debug it, and refine it.
Near the end, you'll prepare a brief presentation explaining the choices you made - what the data said, what you built in response, how you used AI in the process, where you pushed back on what the AI suggested.
Act 3 will get its own detailed page when the week-by-week plan is ready.
What Act 3 teaches
How to scope a real project to fit available time. How to build something more complex than a single page. How to iterate sustainably across weeks. How to defend choices to an audience. How to present work in a way that connects evidence, intent, and output.
What you’ll have at the end of Act 3
- ◆A working tool deployed at your site's Tool page
- ◆A complete three-page site that tells the story of your semester: who you are, what you learned, what you made
- ◆A presentation that explains your work to others
- ◆The experience of having shipped real software that responds to a real community's stated needs
How class actually works
You’ll work on your site every week. Most days will have a short content piece (15-30 minutes - a concept, a demonstration, a discussion) followed by substantial time to work on your pages. Some days will be heavier on content; some days will be almost entirely studio work. But every day, you’ll touch your site in some way.
Content lessons happen when the work calls for them, not on a fixed schedule. When you’re about to need to understand colors, we’ll spend time on colors. When the model is producing something strange and you don’t know why, we’ll spend time on why.
You’ll have an ongoing website observation practice through Act 1 and into Act 2. You’ll have a weekly field notes practice - short notes about technology you noticed in the world that week - across the whole semester. Both are writing practices that strengthen across the course.
A note about the AI in the room
The AI that helps you build your site lives on a computer in this classroom. It’s a real machine you can see. Your work doesn’t go to a company in another state; it goes to a box in this room. That choice is intentional. Your data and your work belong to you and to the school, not to a tech company. Part of what this course teaches is why that matters and how to think about the difference between AI you can see and AI you can’t.
This course is a one-credit, semester-long Computer Science I offering aligned to the Louisiana Student Standards for Computer Science and structured as a deliberate on-ramp to AP Computer Science Principles. The student-facing sections above describe the lived experience of the course; the sections below position it for standards review.
LDOE Computer Science I alignment
Act 1 establishes high-school systems literacy and structured-instruction prerequisites: hardware/software systems (M.CS.1B → H.CS.1A), data representation and tokenization (M.DA.1A → H.DA.1A), and algorithmic decomposition (M.AP.1B → H.AP.1B) - delivered in service of the build, not as standalone units before building can begin.
Act 2 develops data analysis and information-literacy outcomes (H.DA.4B), critique of algorithmic feedback loops (H.IC.2C), and network/cybersecurity reasoning (M.NI.2B → H.NI.1A), grounded in a real school-community survey that students design, administer, and analyze.
Act 3 takes a project through the Software Development Life Cycle (H.AP.4A) with decomposition into components (H.AP.3A) and documentation/communication appropriate to maintenance and debugging (H.AP.4E) - culminating in a deployed tool and a defense of design choices.
AP Computer Science Principles on-ramp
Students completing this course should be well-prepared for AP CSP Big Idea 1 (Creative Development), Big Idea 3 (Algorithms and Programming), and Big Idea 5 (Impact of Computing). The spec-driven workflow that runs across all three acts mirrors the AP Create Performance Task process: specify, build, debug, defend. Students enter AP CSP ready to perform, not just participate.
Vertical articulation
Middle-school computer science core practices map into this course's first six weeks; high-school core practices are developed across all three acts; the course produces students prepared for AP CSP and beyond.
The Sovereign Architect Rubric
Every act artifact is assessed on three pillars aligned to Louisiana standards and AP CSP readiness.
Clarity of Intent
Did the student's spec produce the desired output? Is the writing structured, specific, and testable enough that the AI can act on it?
Auditing Rigor
Did the student identify where the AI guessed, drifted, or failed? Is the verification of AI output documented with evidence?
Sovereignty
Can the student explain why running AI locally is different from using a generic cloud service? Does the build protect community data and respect the school's data boundaries?
The PAR Loop
The Act 3 tool is not a made-up project. Participatory Action Research (PAR) grounds the build in real needs: students start Act 2 by asking what their school community actually needs, and the tool they ship in Act 3 responds to what the data showed. Louisiana's CS integration mandate is satisfied by tools that serve the whole school, not just the CS lab.
Not on this page
- ◆Day-by-day lesson plans (live elsewhere)
- ◆Detailed rubrics for individual assignments (live elsewhere)
- ◆Diverse learner accommodations and modifications (live in a separate document; referenced when ready)
- ◆Assessment cadence and grading scheme (live in the syllabus)