Quest 10 of 15
Core Project: Your AI Study Architect
Apply the first ten modules in a personal study-assistant project using Socratic tutoring, token awareness, optional live search, and clear ethics. This is the core project; modules 11–14 add benchmark evaluation, fine-tuning choices, APIs, and portfolio practice before the final Study Architect showcase in module 15.
Start here
Your final project pulls everything together. You do not need to build an app — you design and demo a study assistant using the skills from Parts 1–9.
Big idea
Your core project combines system prompts, practice habits, cost awareness, ethics, and optional tools into one study assistant. Later modules deepen evaluation, fine-tuning choices, APIs, and portfolio skills before the final showcase.
Learn one idea at a time
Read, explore, then mark each idea when you can explain it.
Idea 1 of 5
Start your capstone with a clear user and explicit non-goals, because these two decisions shape everything downstream. The user should be specific enough to picture: 'Form 4 students revising biology the week before a test', not 'students'. The non-goals — what your assistant will refuse to do — are just as defining: it will not write essays for submission, will not answer live exam questions, will not discuss topics outside its subject. Write these refusals into the system prompt and test that they hold. A project that knows what it refuses demonstrates more sophistication than one that attempts everything, because refusal-by-design is exactly what separates responsible AI products from reckless ones — at every scale from classroom to industry.
Choose a deep dive
Open the topics you want to explore. The detail stays folded until you need it.
Deep dive 1Design for learning, not shortcutting
A helpful assistant asks a learner to attempt a problem, gives a hint, and explains the next step after seeing their work. An unhelpful assistant writes a full answer that can be submitted without understanding. Your design should make the learner think, practise, and explain rather than only receive text.
Deep dive 2A small trust packet
Your project should state who it helps, what subject it covers, what information it must not collect, and when it should tell the learner to check with a teacher. Include two examples of mistakes it could make and how the user should respond. This honest document makes a demo stronger because it shows you understand both value and limits.
Deep dive 3Case study: two capstone demos, one lesson
Two teams demo study assistants. Team A shows a flawless three-minute video and claims the bot 'answers any question in any subject'. Team B shows a live demo that stumbles once, then presents a table: twenty real test questions, fifteen answered well, three partially, two wrongly — with the two failures analysed and a fix proposed for each. The judges score Team B higher. Evaluators, employers, and investors all know that every system fails somewhere; what they are testing is whether the builders know where. Evidence with honest failure analysis beats polish every time.
Deep dive 4Present the system, not just the chatbot
A strong showcase explains the whole design: who the user is, what data enters the system and what is deliberately excluded, which model handles which step, where a human checks the output, and what you measured. Draw it as a simple flow — input, processing, checkpoint, output, feedback — and rehearse answering the three questions every audience asks: What happens when it is wrong? What does it cost to run? What would you build next? Being fluent on those turns a class project into evidence you can use for applications and interviews.
Deep dive 5Capstone: tutor, not cheat sheet
Build a study assistant with Socratic questions, subject limits, and an honest trust packet about data and mistakes.
One-minute challenge
Connect this lesson to real life
Name one situation where this idea could help, and one thing a person should still check.
Explore a real-world example
Use the arrows to connect the idea to a visible situation.
Photo example
Example: demo with classmates
Present the problem, live demo, limits, and what you would improve next — like a mini product pitch.
Key terms
Tap a term to flip and read the definition.
Optional further learningFree textbooks and trusted online resources
These sources informed the course structure. Use them to revisit a concept or study it in more depth.
Ready check
Tick each idea only when you could explain it without looking back.
Ready for practice? Run the Socratic sprint and pitch mentor challenges, then assemble your trust packet for assessment.
Extra context (audience, logistics, curriculum notes)
Built for: Final project suitable for science fair, ICT portfolio, or university application evidence.
Formats: Project studio · Capstone chat lab (full mode) · Presentation day
AI Architect Module 10 — capstone study assistant with system prompt, chat loop, analytics (AI_Architect_10_Part_Course.md).
Next up
Ready for the next part?
When you've finished the reading, inline exercises, and knowledge check for this part, check the box to continue.