
← Course overview · From a moment of care to first real feedback
You can read and print this page without scripts; with scripts on, you can project it page by page.
Care / 00—02 minutes
What would you like to take one more step towards?
First, think of something you care about, then use this case to find a way forward.
Open the teaching points
- People around you, your school, the environment, your interests, or any life you notice can be a starting point.
- It is okay if you are not interested in this app. Just leave with one small step you are willing to take.
Notes for the speaker
You are not required to share family stories or psychological results. Take quiet time to write first; sharing in the session is voluntary. If you are self-studying, you can skip the self-reflection and choose a situation directly.
You do not need a grand plan to begin. Think of something small you have noticed recently, and something small you have managed to do. We will follow the story of Shiguang Huaxia, then return to an idea of your own.
AHALab × ZimaBlueAI · aha-lab.aiLimits of the case / 02—05 minutes
A project that is still finding its way
The course for older adults has not started yet; the app invited a small number of users to try it first.
Open the teaching points
- The course is designed, but it cannot yet be counted as a result that happened in a classroom.
- Someone being willing to try does not mean a market has been found.
- Case studies are closed-source; today we share the process, choices, and methods.
Notes for the speaker
Use only anonymous summaries. Do not share private experiences or post in group chats. First, help students understand that the case is not a success story.
This project is still finding its way. The course about later life has been designed but has not yet been taught. A beta version of the app exists, and a few people have tried it. Its lasting value, market and willingness to pay remain untested.
AHALab × ZimaBlueAI · aha-lab.aiHow it began / 05—08 minutes
Make room for conversation at a person’s own pace
You can start with a broad direction, then find one specific small thing to try.
Open the teaching points
- Concern: How can we create one more place where people feel ready to speak?
- Need hypothesis: You can speak without writing an article first, and you decide when to pause or share.
- Step 1: Ask—Answer—Pause—Keep a record.
Notes for the speaker
The answer is no. Older people and their families may have different expectations, so ask each separately. Do not portray all older people as lacking company.
Concern about conversation was the starting point, but making an app meant choosing a smaller question. First, make one round of asking, speaking, pausing and saving work smoothly. Then consider whether to continue. Being willing to talk does not mean being willing to let someone else read it.
AHALab × ZimaBlueAI · aha-lab.aiFrom a problem to a need / 08—11 minutes
Separate the complaint, the need and the feature
Starting with first principles: Separate goals, facts, constraints, and assumptions.
Open the teaching points
- Observation: Some people are not used to typing for long periods; you still need to ask about specific people.
- Need hypothesis: When you want to speak, you do not need to organize it into an article first, while keeping consent and data boundaries.
- Question “must make an app”: Paper cards, human interviews, and voice can be tried first.
Notes for the speaker
Within the original 3 minutes, practice breaking down assumptions once; do not start a separate concept lecture. “Expression autonomy” is a value choice, not a physical law. Check if the solution truly serves the need; experience is useful, but facts must be verified. See the Self-discipline and First Principles page.
Saying 'make an app' already chooses a solution. Take one step back: what is happening, what is getting in the person's way, and what would they like to change? Separating facts, constraints and guesses may reveal a much smaller way to help.
AHALab × ZimaBlueAI · aha-lab.aiActivity 1 / 11—16 minutes
Turn what you noticed into a question you can test
Who × Situation × Barrier × Desired change.
Open the teaching points
- Write the phenomenon first, then turn it into one need statement.
- Circle one guess and write down evidence that might disprove it.
- When studying alone, check your work using the prompts. If you have a voluntary partner, you can also review each other’s work.
Notes for the speaker
When self-studying, write first and then look at the prompts. In meetings, allow quiet time for writing and voluntary sharing. The alternative teaching scenario is fictional: forgetting to return borrowed items. Do not treat this exercise as real user needs.
Write down what you think the person needs. Circle anything you do not yet have evidence for. You do not need to wait for classmates to be online. Question the idea from another angle yourself, or later invite someone who wants to take part.
AHALab × ZimaBlueAI · aha-lab.aiGuiding choices / 16—19 minutes
What you protect shapes the choices you make
When sharing layouts are very appealing, why do you still want to start with a slow chat?
Open the teaching points
- Protect: current expression, your own pace, and sharing choices.
- Pause for now: adding a complex flow just for a quick demo.
- This is the current choice; you can reconsider if new evidence appears.
Notes for the speaker
You can choose to prioritize expression pace while development validation continues quickly. The book feature already exists, and the output route is still being considered. Do not infer from this choice that sharing itself is harmful.
During development, Shiguang Huaxia faced a choice: add more sharing and page-design features, or first protect the pace of conversation? For now, the choice is to let people talk slowly. That is a decision by the creator, and new evidence may change it.
AHALab × ZimaBlueAI · aha-lab.aiDrawing on experience / 19—22 minutes
Build on what you already know
Apply experience, resources, and tools to the task at hand.
Open the teaching points
- Existing voice skills, practical experience, and personal strengths can all connect to the current task.
- AI can suggest solutions, write prototypes, and help find errors.
- Information gaps still require finding sources, testing personally, and reporting results.
Notes for the speaker
The case was built quickly, supported by past accumulation. Self-learners list their available resources and gaps, then plan the next test. There is no requirement to contact an expert immediately.
A quick first version often draws on things learned long before it. Experience with speech tools, people you can ask, documents you can check, and your own practice in listening, organising or making things can all help with the next step.
AHALab × ZimaBlueAI · aha-lab.aiBuilding with AI / 22—26 minutes
You can start before you know platform programming
First, clarify what this session covers and how to check it when done.
Open the teaching points
- Tell AI: goals, known facts, real constraints, hypotheses to test, and verification criteria.
- Ask AI to list status, unknowns, and testing methods before providing implementation.
- Use synthetic content to complete one start–pause–save cycle yourself.
Notes for the speaker
Check the teaching prompts in the AI and tool pages. You can draw a state diagram first; you do not need to finish the entire App on the spot. Remember to cover permissions, APIs, data, and delivery. You are responsible for verification.
You can start a prototype with AI before learning all of platform programming. You still need to explain the task, its limits and how you will check it. Try the version AI produces on the device yourself.
AHALab × ZimaBlueAI · aha-lab.aiMeeting difficulties / 26—29 minutes
Progress comes from trying it again
After it runs, there are still choices for playback, permissions, returning, and recording.
Open the teaching points
- Users hear the question first, instead of facing a blank screen.
- Open the camera for a preview first, then confirm before recording.
- After opening the reading page, make sure there is a way back.

Notes for the speaker
This is a process summary. It does not claim one prompt solves all problems. Each step states only expected vs. actual results, without guessing unverified causes.
Real use reveals small obstacles: a question is not spoken aloud, opening the camera starts recording at once, or there is no obvious way back after reading. Such details can affect whether a person wants to continue. They need to be tried again and again.
AHALab × ZimaBlueAI · aha-lab.aiActivity 2 / 29—33 minutes
AI says it is fixed. The phone is still silent.
Known → Unknown → Minimal check → Single change → Re-test.
Open the teaching points
- Write down the expected and actual results, not just “it doesn’t work.”
- Give a testable hypothesis, then design one minimal check.
- Keep evidence of failure; do not change everything at once.
Notes for the speaker
Write down what you know, what you don’t know, and the minimal check, then open the self-check prompts. In meetings, you can share anonymous synthetic examples. Volume, triggers, and service errors are all hypotheses to check.
Treat 'no sound' as an observation first. Volume, the playback trigger and a service response are possible causes. Another instruction to 'fix it' may not help as much as one piece of evidence that rules out a guess.
AHALab × ZimaBlueAI · aha-lab.aiSafety and responsibility / 33—37 minutes
A usable prototype still comes with responsibilities
Someone is responsible for records, authorization, and call costs.
Open the teaching points
- Run through it with synthetic data: where is it saved, who can access it, and can it be restored?
- A personal Key is for authorization and quota; data isolation needs separate controls.
- The client still has risks of sensitive settings being extracted. Keeping the source code closed does not guarantee safety.
Notes for the speaker
The beta has remaining risks. We do not present it as a production-ready solution. We only describe a simplified data flow. We do not show configurations, allocation pools, or keys. University groups may add cost limits. We do not describe attack steps.
Once someone can try the app, you still need to know where information goes, who can access it and who pays. A personal API key can help manage access and limits, but keeping people's records separate requires other checks. A beta can still carry risks.
AHALab × ZimaBlueAI · aha-lab.aiA small trial / 37—40 minutes
Sending the installer is only part of delivery
Invite → Install → Complete a small task → Follow up.
Open the teaching points
- Willingness to try is an attitude signal.
- The install file and entry point may be the first barrier.
- Ask the person to complete only one voluntary small task at a time.
Notes for the speaker
The classroom does not distribute closed-source apps. Use an anonymous text summary as a demo. Success criteria include finding the pause button on their own. Do not set a required speaking duration for older adults.
Sending an installation file does not mean the person has installed it, or knows where to begin. Invitation, installation, completing a small task and a follow-up are different things to observe.
AHALab × ZimaBlueAI · aha-lab.aiFeedback / 40—44 minutes
Ask what happened before adding a feature
First ask what is not working: the task was not solved, the entry point was not found, or it could not be used at all.
Open the teaching points
- Want to change the text but cannot find where: check if the entry point is hard to find.
- Want a new topic, then find an existing option: check if the prompt is clear and the option is easy to see.
- Different recording habits: maybe you need to rethink the user scenario.
Notes for the speaker
Only provide anonymous rewrites; do not read testers' private matters. Next-day feedback is not long-term retention; a psychology background is not a professional efficacy endorsement.
A request for a feature may point to an unmet need, a hard-to-find control, or trouble installing or using the app. Ask about the most recent experience before deciding what to change.
AHALab × ZimaBlueAI · aha-lab.aiActivity 3 / 44—47 minutes
When tools can copy, what does a creator still decide?
Learn the method, check permissions, use your own assets, and take responsibility for the results.
Open the teaching points
- Use REA and replica-skill as read-only observation material. Do not install or run them.
- A tool license does not equal permission for target code, brands, or user data.
- Similar looks still require you to explain your users, choices, and verification.
Notes for the speaker
First, independently choose one usage method. Then explain the permission, source, and what you need to verify. Share voluntarily in the meeting. Use read-only tool materials. Do not install or analyze third-party apps.
When tools can analyse and recreate things, someone still has to decide which materials to use, who the work is for, what to protect and whether the result works. A tool's licence does not give permission to use someone else's code, brand or data.
AHALab × ZimaBlueAI · aha-lab.aiAction and setbacks / 47—50 minutes
Make room for small commitments, and for starting again
Set a small goal, reduce one distraction, and reschedule after an interruption.
Open the teaching points
- Set time and place, start prompts, and completion evidence. Keep rest and cost limits.
- If it fails, bring back the phenomenon. If it does not start, check the blocker. Ask for help or re-agree in time.
- Self-discipline helps execution; first principles help check direction. Neither guarantees success.
Notes for the speaker
3 minutes still include failure and responsibility. For example, “If I sit at my desk tomorrow night, I will draw two paper cards first.” Do not imitate famous people staying up late. Allow pausing with a reason. Do not blame failure on personality. Do not prove the market with wealth or likes from friends and family. The methods page has two extension images.
Once you have chosen a small goal that matters to you, make starting easier. Put the materials ready, reduce one distraction, and set another time if you are interrupted. If the need turns out not to be there, allow yourself to stop for a reason.
AHALab × ZimaBlueAI · aha-lab.aiActivity 4 / 50—57 minutes
Keep a wider hope. Choose a smaller next step.
Use one of your strengths, check one guess, and plan a small, doable try.
Open the teaching points
- 2 minutes to write the target and the guess, 2 minutes to write the action and the evidence.
- 2 minutes to self-check using the prompts, 1 minute to revise; if you have a partner, you can choose to share.
- Use one existing strength, and write down the start conditions and what to do if you get stuck.
Notes for the speaker
Keep it to about 7 minutes. Focus on the target, action, time, evidence, and plan for sticking points. You do not need to fill in every box. University students work on their own and can ask for help online. If you have no topic yet, just observe. Respect it if the person you ask says no.
Now return to your own idea. Connect the person or situation, the assumption, the action, the time and the evidence. Fill in only what this attempt needs. Without a partner, use the self-check questions; you can ask for advice online later.
AHALab × ZimaBlueAI · aha-lab.aiTake one action away / 57—60 minutes
Next time, bring back something that happened
Use one action and the feedback to keep adjusting your idea.
Open the teaching points
- Finally, write three sentences: what I care about, when I will act, and what I will bring back next time.
- Check if it started after 48 hours. Check evidence and revisions after 7 days.
- Start with what you care about. Use tools to help, and consider how your choices affect others.
Notes for the speaker
Self-learners keep their own cards. If you need advice, manually send a non-sensitive summary you are willing to share. Sharing in meetings is voluntary. Project success is not judged by in-class statements or reading marks.
The lesson pauses here; the project is about to meet everyday life. You do not need to have its long-term direction settled. Let one small action happen first. Come back after seven days with what you noticed, then decide whether to continue, change or pause.
AHALab × ZimaBlueAI · aha-lab.ai