Workshop kit · FTSM Technopreneurship Bootcamp 2026 · Day 2 · Build
Everything for the 2-hour session in one place: the minute-by-minute plan, the thinking lenses as live visuals, the scaffold groups work through, the prompts, the scoring rubric and the prep list.
01 · Run sheet
Drag the clock or press play to rehearse it. The panel shows which slide is up and whether an AI build is running at that moment, so every wait is planned work.
02 · Concepts
Each lens is taught in about two minutes and then used straight away on the Thinking Canvas. The chain is: system → user → hypothesis → slice → prompt → test → reflect. Each visual plays as you scroll to it; press Replay to watch it again.
The map
Day 1 and this morning covered the problem space. This session is Develop: explore solutions by prototyping, then converge on one MVP. If the peer test shows the wrong problem was defined, groups go back to the first diamond. That is normal, not failure.
Design Council (UK), 2004 · Framework for Innovation
What “viable” means
A prototype asks whether people understand it; an MVP asks whether they will use or pay for it; a product asks whether it scales. The MVP is one core job that works, does not break in the demo, needs no help, beats the old way, feels good and matters. That is why the canvas allows only three features.
After S. P. Anderson’s UX hierarchy of needs (2011), building on A. Walter (2011); MVP-slice view as drawn by B. Waldock, beingagile.co.uk
How an MVP is born
Design thinking climbs from needs to a concept. Lean UX turns the concept into a hypothesis and the smallest thing that tests it. Agile takes over with short build–measure–learn loops: Sprint 1, the peer test and Sprint 2. AI shrinks each Build step to minutes.
Build–Measure–Learn: E. Ries, The Lean Startup (2011)
Lens 1 · Canvas box 1
Ask why the problem keeps happening and who else is part of it. An app that only reacts to events is weak; one that changes a structure (a shared live schedule) is strong. The people in the system, such as lab staff, the faculty office and other groups, decide whether the MVP can work.
Box 1 → LOOK & DATA in the promptTry it · build a point of view
Lens 2 · Canvas box 2
Start from a person and an insight, not a feature list. The point-of-view sentence forces three things: a specific user, a real need (a verb, not a feature) and an insight into why it is hard today. The “How might we…?” question turns it into an open design challenge the AI build answers.
Empathise · Define · Ideate · Prototype · Test, after the Stanford d.school
Box 2 → ROLE & GOAL in the promptLens 3 · Canvas box 4
Think like a computer before you prompt one. Decompose the core job into steps, spot a familiar pattern, abstract away everything that does not test the hypothesis, and write the rule precisely. Each habit fills one part of the build prompt.
J. M. Wing (2006), Communications of the ACM 49(3)
Box 4 → CORE FLOW + FEATURESLens 4 · Canvas box 3
Every idea hides one assumption that can kill it. Write it as a hypothesis a test can prove wrong: who, what they will do, why, and the signal you will see. The signal must be visible in today’s two-minute peer test.
Format adapted from J. Gothelf & J. Seiden, Lean UX (2013)
Box 3 → the peer test at 11.5003 · Generation-time playbook
A first build takes 2–5 minutes and each tweak 1–3 minutes. Run a simulated build to see what every role does while the AI works, then check the fixes for builds that misbehave.
Writes the next prompt: one change only.
Writes 2–3 test steps for this build.
Drafts the 20-second demo script; logs prompts.
Prepares sample data and app wording (RM, kolej, BM/EN).
Wait one more minute, then refresh and resubmit a shorter prompt.
Paste the same prompt into a backup tool: Lovable or Bolt.new.
Undo to the last working version and split the change into smaller prompts.
The next member signs in with their own account and continues from the saved prompts.
Undo it and demo the working version. A working simple app beats a broken clever one.
04 · Thinking Canvas
The canvas shows only the question. A stuck group gets the Level 2 questions first, then the Level 1 starter and example. Pick a box and step through the levels the way a mentor would.
05 · Prompt cards
Each bracket comes from a canvas box. Edit the fields and the four-part build prompt assembles live; box 3 is not prompted, it becomes the peer test. Then iterate with one change per prompt.
06 · Peer test & demo verdict
At 11.50 a tester from another group gets two minutes and no explanation. At 12.10 mentors score each speed demo on four criteria. Tap the scores to rehearse a verdict.
Peer test · 11.50 – 11.56
Reflect · 12.25
System: what did the tester show us? User: did they behave as we believed? Hypothesis: confirmed or not, and why? Next: diverge again, or converge and build the next slice?
Speed-demo verdict · 0–2 each
07 · Prep
Ticks are remembered in this browser only. Send the student message through the organisers before departure.
Message to students
FTSM Technopreneurship Bootcamp – Day 2 "AI-Powered Prototyping" (Dr Akmal) Please prepare BEFORE leaving for Tasik Chini: 1. Bring a laptop + charger (phones alone are not enough). 2. Sign in once at aistudio.google.com with your Google account and open "Build". 3. Create a free account at lovable.dev (backup tool). 4. Bring your group's problem statement from Day 1. Internet at the site may be limited – one phone hotspot per group is a good backup.
Plan B · no internet
Play the recorded demo. Groups sketch four phone screens on A4 and run the peer test on paper. They build the AI version tonight, 9.00–10.30 pm, and mentors check it at the 9.15 am pitch practice.
08 · Downloads & links
Print the Thinking Canvas on A3 landscape and everything else on A4. Groups can work on paper or in MVP Studio, the interactive version of the same scaffold.