How do you communicate that empathy to another person?
Empathize through visual storytelling
Customer Journey Mapping (CJM) is the tool we'll use to collect and tell the story of our users
Kid Mobile Case Study
The kid's persona
Name: Akanksha
Gender: Female
Location: Pune, India
Age: 10
Description: Lives in Pune with her elder brother and parents. Loves chocolates, aspires to be a
badminton player, hates assignments and exams. Owns a tablet, would like a pet.
Customer Journey Map — Before
CJM — During
CJM — After
List of problem statements
The kid plays on the mobile for most of her spare time
She snacks/eats while playing on the mobile
She spends very little time playing outdoors
Customer Journey Mapping
Write the customer's activity
Write the major steps of the user experience
Split the story into 3 — Before, During and After
Mark the emotional highs and lows (😄 and 😞)
Say the pain as a test — Given / When / Then
The CJM in one testable line — current behaviour, no solutions:
Given Akanksha has free time and a tablet,
When she reaches for something to do,
Then she plays indoors for hours and skips going out 😞
This is your Red — the pain you'll later prove you fixed. Still no solutions here.
All three pains — each becomes a Red test
Same move for every pain on your CJM — current behaviour, no solutions:
Given free time + tablet in reach, When she looks for something to do, Then she defaults to the tablet for hours 😞
Given she's deep in a game, When she feels like eating, Then she grazes at the screen, not noticing how much 😞
Given a free afternoon + good weather, When she picks what to do, Then she stays indoors and skips badminton 😞
Each asserts a desired outcome that isn't happening yet — that's why it's Red.
What a Red test is for
Make it falsifiable — a pain you can't fail is an opinion, not a test
Run it on your 3 real customers — does the real person actually fail it? Pain confirmed (Red is real). If they already pass it, it's not their pain — drop or rewrite it
That's the anti-fabrication rule in test form: a Red test proves you didn't invent the problem 🚫🤖
Green isn't now. You turn it Green only in Test / Iterate — re-run the same scenario on the same customer and the Then lands on 😊 (she picks up the racket). Proven by the customer, not by AI.
Session 2 - Work Out
A timed working block. Pen & paper / your own doc — AI comes in S4.
Lock your project with your buddies — say it out loud as a How Might We (15 min)
Hypothesis CJM — sketch your customer's Before · During · After from what you guess today (30 min)
Write your Red tests — Given / When / Then, current behaviour, no solutions
Buddy swap — attack each other's tests: could the real person already pass this? (15 min)
This is a hypothesis, not the truth — carry it to the field as your interview checklist: guesses to confirm or kill. 🚫🤖
For each hypothesis from Session 2, write the question or observation that could disconfirm it — never name the answer you want:
Hypothesis (a Red test)
Disconfirming probe — your field instrument
She defaults to the tablet for hours
"Walk me through yesterday after school — what first, then next?"
She snacks nonstop, unaware
Watch a play session; after, ask "what did she eat today?"
She stays indoors, skips badminton
"Tell me about last Saturday afternoon." (don't mention badminton)
That one page is your interview guide.
AI move (pre-field): ask AI to draft an interview guide from these hypotheses, then "which of my questions are leading? rewrite them neutral."
Dry-run — rehearse the interview
Don't let your first interview be with your precious real customer. Rehearse today:
Interview your studio buddy for 10 min — use your field guide
Buddy gives feedback on leading questions + the follow-ups you missed
Swap and repeat
Tighten your guide from what you learned
Now your first real interview won't be your first interview ever. 🎯
Launch — your one piece of real "homework"
Over the Jun 19–21 weekend, go out and meet your 3 customers (more is better)
Your studio buddy comes along — company, courage, and a second set of notes
Bring back: notes, photos/video (with permission), and direct quotes
We open the Empathize showcase the moment we're back, Mon Jun 22
Task 1A: Empathize This (1/2)
Interview guide
The one-pager you built today — each hypothesis paired with a disconfirming probe (your field instrument)
Interview guide
Note which of your questions were leading and how you made them neutral
Field work
Interact with at least 3 customers (more is better)
Field work
Anyone not from your UP class; age ±15 years from yours
Field work
Bring back notes, photos/video (with permission), and direct quotes
Task 1A: Empathize This (2/2)
Create a presentation with the following:
CJM
Create personas (name, gender, age, location, occupation, family)
CJM
Write the customer activity
CJM
Before / During / After, with emotions and photos of the customer
Problems
List the problem statements — no solutions, please
Task 1B: Empathize That
Create a video:
A short video (2–5 min) — a skit highlighting the customer's current status
Use human actors (no AI/avatars/cartoons)
Minimize dialog; use your CJM as the script
Upload the video link with your portfolio
Checklist for Empathize
[ ] Interview guide — disconfirming probe per hypothesis, leading questions fixed
[ ] Customers — at least 3
[ ] CJM — persona(s) with name, gender, age, location
[ ] CJM — main activity of the customer
[ ] CJM — 3 sections: Before, During, After
[ ] CJM — emotions (smileys/sadeys present)
[ ] Problems — listed
[ ] Problems — NOT solutions
[ ] Uploaded: presentation + video
Refactor — sharpen your 3 questions
Reopen the portal. Rewrite the 3 questions you wrote cold this morning.
Which were leading? Make them neutral.
Which fish for an opinion? Turn one into "tell me about the last time…"
The gap between your Red and your Refactor is today's learning. 🎯
Session 4 · Empathize — Showcase
By the end of Session 4 — you'll walk out with…
1
A real-customer CJM, AI-synthesized & verified
2
Your story presented in your room
3
Phase-1 (Empathize) portfolio submitted
Back from the field — turn raw notes into a CJM
You've met real people. Now — and only now — AI earns its place: it organizes what you actually heard.
AI move — Empathize after you've met real people
First, you met your 3 humans. Now paste your raw notes/transcripts → ask AI to draft a CJM (Before/During/After) and cluster pains into themes
Ask it to scaffold a persona — then fix every field against the real person you met
⚠️ If AI invents a quote, a name, or a "typical user" you never met — delete it. That's fabrication.
⚠️ AI gives you the average customer; your marks come from the specific one. Verify, don't trust.
Cite it: one line — "AI clustered my interview notes; personas verified against real customers."
Worked example — notes → prompt → AI draft
Your raw notes (one real customer): Akanksha, 10, Pune. Home from school → tablet games for hours. Snacks while playing. Rarely goes out. Loves badminton but "too tired". Parents frustrated
Paste them into AI with this prompt:
You're my Design Thinking assistant for the Empathize step.
Below are my RAW notes from ONE real customer I interviewed.
1) Draft a Customer Journey Map — Before / During / After, with an emotion (happy/sad) at a step that has emotions.
2) List problem statements — the PAIN only, NO solutions.
3) Flag anything you INFERRED that is NOT in my notes, so I can verify.
Rules: do not invent quotes, names, or facts. Use only my notes.
NOTES: <paste your notes here>
AI draft you then verify:
Before tablet after school — excited 😊 During hours of play + snacks, skips outdoors — absorbed → restless 😞 After parents nag, homework undone — tension 😞
Problems:
free time defaults to screen
no easy off-ramp to go outside
no parental guardrail
⚠️ Inferred — verify with her: "too tired for badminton"
Showcase format
In your room of 5: 4 min present + 2 min peer feedback — everyone, every phase
Main stage (the roulette champions): 6 min present + 3 min Q&A
Audience questions earn Live-Engagement XP
Empathize Roulette — pick the room champions
Presentation focus this phase — Story
Open with one customer, by name, in one human moment
Walk us through the journey — let the 😄 and 😞 do the work
No solutions yet — leave us feeling the problem
Criteria — Empathize portfolio (/10)
Checklist completeness (50%)
Video (15%)
Subjective quality (20%)
Customer evidence (15%)
How the room works each phase
Breakout presentations — 6 rooms of 5; everyone presents 4 min + 2 min peer feedback
Peer-review round — Rank your roommates on the checklist (earns your peer-review marks)
Roulette to the main stage — the wheel sends one champion per room to the whole class
Everyone uploads their own portfolio to Shared Drive; the AI grades all of them
Why 1: Why does the kid play on the mobile for such a long time?
Ans: To gain experience (XP) points.
Why 2: Why does the kid want to gain XP points?
Ans: These are rewards that the kid can brag about to friends.
Case Study: Application of Multi-Whys
Why 3: Why does the mobile app reward the kid for experience?
Ans: Rewards keep the kid addicted to playing the game.
Why 4: Why are children exposed to such addictive mobile apps?
Ans: A lack of control mechanisms exposes children to all apps.
From symptom-Red to root-Red — the baton from Empathize
Your Red test names a symptom. Build at the symptom → you get a band-aid (the "screen-time lock app" everyone leaps to).
Multi-Why drills the Then of the test:
Then she stays indoors → why? badminton needs a partner → why? nobody free at short notice → why? no way to see who's around right now → root: no low-friction way to find a nearby playmate on demand
Now rewrite the test at the root:
Given a free afternoon, When she wants to play badminton, Then she finds a partner within 10 min
That root-level test — not the symptom — is the one you carry into Solve and try to turn Green.
Task — Analyze This
Add to your presentation:
Take the problems from the Empathize stage
Apply Multi-Why analysis on all of them
Tip — No solutions, please
Chain the whys; branch where it makes sense
Analyze This — Checklist
[ ] Chain of why statements rather than independent whys
If a thing happens ≈ Given/When a situation occurs
Then a good consequence ≈ Then the outcome you want
But a bad consequence = the tension your solution must resolve
Your HMW is just the Then you're aiming for — your future acceptance test.
Put on your movie critic's hat
Think of a movie or novel you really liked
Pick a scene you liked
Find the conflict the character faced
Represent it in the model (If… Then… But…)
Time left:
Fine Formulation — Zone & Time of Conflict
Zone — the space where the conflict occurs (e.g. 3–5 cm around the tyres)
Time — the moment the conflict occurs (e.g. 150 ms before impact)
Conflict of Interest — Kids Mobile Case Study
If
kids are given access to too many apps
then
they have more freedom to discover new apps
but
there's no parental control over the type of apps
So the problem becomes — How might we give more freedom to kids while retaining parental control?
Zone — around the "Install" button · Time — at the moment of installation
Task — Analyze That
Add to your existing presentation:
Take the problems from the Multi-Why stage
Mark the level of the problem you'd like to solve
For each, ask "What would you like to improve?" and "What stops you?"
Frame each as a conflict (If… Then… But…)
Write each desired result as a How Might We question
Note the zone and time of conflict
Analyze — Checklist
[ ] The key Whys/problems from Multi-Why
[ ] Conflicts defined for each (If/Then/But)
[ ] Positive and negative consequences stated
[ ] Desired results as How-Might-We questions
[ ] Zones and times of conflict
AI move — Analyze with a skeptic at your elbow
Feed your problem → ask AI to push the Multi-Why deeper ("ask why 3 more times; branch where it splits")
Ask it to propose If/Then/But conflicts and map them to TRIZ-40 parameters
⚠️ AI loves a tidy, plausible, wrong root cause. Every why is a hypothesis — check it against what your customer actually said.
⚠️ A why-chain that reads beautifully but contradicts your CJM is a red flag, not a win.
Best use: AI gives you more whys and conflicts; you keep only the ones the real customer would recognise.
Worked example — problem → prompt → Multi-Why + conflict
Your input (one problem from Empathize): "Akanksha defaults her free time to tablet games and skips going outside."
Prompt to the AI:
You're my Design Thinking assistant for the Analyze step.
Here is ONE customer problem from my real field work: <paste it>
1) Build a Multi-Why chain (Toyota 5-Whys) — each why caused by the next;
branch if it splits. NO solutions.
2) From the deepest why, frame ONE conflict as If... Then... But...
3) Turn it into a How-Might-We question, and name the zone + time of conflict.
4) Flag any why you INFERRED that my problem statement does NOT support,
so I can verify it with her. Do not invent facts.
AI draft you then verify:Why play? → instant reward · Why want it? → brag to friends · Why brag? → wants to belong · Why screen over outdoors? → screen is frictionless, outside needs setup.Conflict — If we cut tablet time, Then she goes outside more, But she loses the social reward she values. HMW — How might we give her the social reward without endless screen time? Zone/time — the first 10 min after she gets home. ⚠️ Inferred — verify with her: the "wants to belong" why.
Refactor — sharpen your conflict
From your own login: post your sharpest If/Then/But and the HMW it produces.
Compare it to your rough Red attempt this morning. Sharper? That delta is today's learning. 🎯
Before you build anything, write 2–3 acceptance tests your prototype must pass:
Given a parent sets up the tablet, When the kid opens a new app, Then it needs a parent OK
Plain language, customer's point of view. If you can't test it, it isn't done.
This is your Red. Next you build only enough to turn these Green.
AI move — your tests become the build spec
Brainstorm co-pilot: race the AI for volume — a generic idea is worth nothing; cross it with TRIZ + your customer
TRIZ helper: ask AI to map your conflict to the 40 principles, then sanity-check on https://triz40.com
Tests = build spec 🚀: your 2–3 Given/When/Then tests are your build checklist — and the executable spec you hand to Antigravity/Claude Code: "Build the simplest app that passes these." Write them first; the agent builds only enough to turn them green. Red → Green, for real.
Worked example — HMW → prompt → ideas + tests
Your input (from Analyze): "How might we give Akanksha the social reward without endless screen time?"
Prompt to the AI:
You're my Design Thinking assistant for the Solve step.
My How-Might-We: <paste it>. My conflict: <paste If/Then/But>.
1) 8 ideas — tag any TRIZ principle each one uses.
2) Consolidate the best into ONE concept (one paragraph).
3) 2-3 acceptance tests (Given/When/Then) — a STRANGER could observe
the Then; do NOT name my solution; plain customer language.
4) Assumptions I must verify with a real customer. Invent nothing.
AI draft you then verify:Ideas: outdoor "social streak", friend-match after play, parent-set play budget… (TRIZ 1 Segmentation, 10 Prior Action). Concept: a gentle screen budget that unlocks a friend meet-up outdoors. Test —G Akanksha has free time at home, W her usual play session ends, T within 15 min she's outside, active, with a friend. ⚠️ Verify: that she values an outdoor social reward at all.
Task — Write your tests
Bring the one concept you built last session (Session 9)
Turn your root-cause problems (from Analyze) into 2–3 green targets — flip the Then, make it measurable
Lock those green targets as your 2–3 acceptance tests (Given/When/Then) — before building. You don't move them afterwards.
Task — Solve That (investor pitch)
For the next round with your "Investor", create a new presentation (~5 minutes):
Empathize — a simple customer story
Analyze — your main problem/conflict
Solution — the USP of your solution (+ a short feature list)
Improvements / advantages (think metrics, numbers) — for user and investor
Plan of action (prototyping + milestones with time)
Support you need from the investor (network, funding, people)
Solve — Checklist
[ ] Green targets set (Red Then flipped, measurable, root-level)
[ ] Ideas from silent brainstorming (8+ good, 12+ excellent), each tagged to a problem
[ ] Ideas from TRIZ methodology (conflict drawn from a root-cause problem)
[ ] Consolidated concept
[ ] 2–3 acceptance tests (Given/When/Then) written before prototyping
[ ] New investor presentation
Criteria — Solve portfolio (/10)
Silent brainstorming volume (15%)
TRIZ genuinely applied (20%)
Novelty of the idea (20%)
Prototypability (15%)
Acceptance tests (Given/When/Then) before prototype (15%)
Consolidated concept (15%)
Bullet Proofing
Rules for the Presenter
Present your concept to a small group (one-way, ~5 min)
Note all feedback in your notebook (one-way, ~3 min)
Session 11 · Test — Prototyping & AI-assisted Build
By the end of Session 11 — you'll walk out with…
1
A working prototype aimed at your green targets
2
Your acceptance tests ready to run on customers
Prototypes
In this step
Build a very basic prototype
Take your prototype to the user and observe how they react
Photograph/videograph their reactions (with permission)
List the problems the customer faces with this new prototype (Test.v1 / Empathize.v2)
Run each acceptance test on a real customer — mark it PASS or FAIL
Physical prototype? Use a lab
Physical prototype? Book a lab for the Thu Jul 09 workshop day or the Jul 10–12 weekend
Take your studio buddy — safer and faster with two pairs of hands
For apps: Figma / Draftbit. For services: roleplay + flow diagrams.
Low-fidelity is fine — clay, cardboard, wires, things lying around
Build it to pass your acceptance tests, then test on a real customer before the Mon Jul 13 showcase
AI move — build path 🚀: ship a real app
Point Antigravity / Claude Code / Lovable at your Given/When/Then → it builds a working app
Push to GitHub → deploy (GitHub Pages / Vercel) → you now have a live URL
Real customers use the live app, not a mockup
🟢 Green = a real customer makes the Then happen on the live app
🔴 NOT green = "the agent says all tests pass" or "the build deployed". The AI is not your customer.
Your git commit history is your iteration log — V1 → V2 is right there in the diff. 🏆 Ship It badge.
AI move — partner path: AI helps you make & validate
AI drafts your Figma flow, your test script
But a synthetic / simulated "user" is not a real customer — every PASS/FAIL needs a human
Whichever path you took: run each acceptance test on a real person, log PASS/FAIL with evidence. 📸
Worked example — failing test → prompt → commit
Your input (a red test from V1): "Given the daily limit is reached, When Akanksha keeps tapping, Then the app locks and shows the meet-up prompt" — it didn't lock.
Prompt to your build agent (Antigravity / Claude Code):
Here is a FAILING acceptance test for my app (repo is open):
<paste the Given/When/Then>.
1) Make the SMALLEST change that turns THIS test green. No new features.
2) Show me the diff + explain it in 2 lines.
3) Commit with a message that references the test.
Do NOT mark it passing yourself — I test it on a real customer.
Agent output:Plan: lock the game view when timeUsed ≥ dailyLimit. Diff: + LockScreen.jsx, guard in App.jsx. Commit:lock games at daily limit (GWT-3). → You redeploy, hand it to Akanksha, mark PASS/FAIL. ⚠️ The agent saying "tests pass" is not green — Akanksha making the Then happen is. The commit is your iteration evidence.
- Emacs - this old editor rocks! - https://www.gnu.org/software/emacs/
- Spacemacs - this new configuration is space age - https://spacemacs.org
- Org Mode - I run my life using this - https://orgmode.org
- Reveal.js - cool presentation script - https://revealjs.com/
- Org-reveal package - lets me live in org-mode - https://gitlab.com/oer/org-re-reveal
- Org-teaching - original codebase for this presentation - https://gitlab.com/olberger/org-teaching
- Plantuml - for all the cartoon work - https://plantuml.com/
- Hugo - for converting into static html - https://gohugo.io
- Gitlab - for hosting my website and the presentations - https://gitlab.com
Acknowledgments
Conversations with
Dr. Dmitry Kucharavy, Professor, France
Dr. Murali Loganathan, co-Founder, Head, Sales & Research, Spirelia Innovation