Capstone

Ship something real

Solve one real problem from your own life or work using at least three capabilities from the week.

Use at least three of these

The point is to combine what you learned, not to use everything. Pick the three (or more) that fit your problem.

A well-engineered prompt or custom system prompt

A multimodal element (image, video, audio, or speech-to-text)

A RAG / personal knowledge base

An agent or automation flow

Multi-language support

A privacy/security measure (safe key handling or a locally self-hosted model)

Deliverable

A short working demo plus a one-page write-up: what problem, which tools, what you learned, what you'd do next.

Example projects

·A 'chat with my documents' research assistant
·A multilingual customer-reply automation
·A private local assistant that never touches the cloud
·An AI-narrated explainer video
·An automated weekly-report generator

How your peers evaluate it

Each peer scores your project 1–5 on every criterion and leaves one written strength and one written suggestion. Remember the values: criticise in private, praise in public.

Criterion	What peers look for	Score
Clarity of problem	Is the real-world problem clearly stated?	1–5
Right tool for the job	Were the AI capabilities well chosen for the task?	1–5
Correct / verified output	Was the output checked, not just accepted?	1–5
Privacy & safety handled	Were credentials and sensitive data handled well?	1–5
Creativity	Is the approach original or especially resourceful?	1–5
Presentation	Was the demo and write-up clear and compelling?	1–5

PreviousDay 3 · Privacy FinishBack to overview