Nudge V1. Small enough to ship, honest enough to learn from.
My read of your brief, the one question the first version has to answer, the experience I'd build to answer it, and exactly what I'd leave for later.
Prepared by Kshitij, Shuru Technologies · October 2026
Draft for discussion
Press → or scroll to begin, ? for keys
What I heard
People don't need more advice. They need a way to notice how they show up, one small nudge at a time.
Nudge is
A calm, personal, intelligent daily loop. You choose how you want to show up, reflect on something real, and get back a response that feels like it was written by someone who has been paying attention. Over weeks, it helps you see your own patterns.
VoiceClarityEmotional awarenessEthical presenceCourage in action
The five pillars are the frame behind every prompt, not a menu the user sees.
Nudge is not
A course platform
A journaling app
A generic AI chatbot
A habit tracker with streaks and scores
A social network
I'm holding myself to this list. Every time a feature idea comes up, this is the first filter.
The one question V1 must answer
Can Nudge create a growth experience people come back to on their own?
Not because a streak is at risk. Not because a notification pinged. Because yesterday's response was worth having, and today's might be too.
Everything in this proposal is in service of answering that one question with real users, as quickly and cheaply as I can.
So the first version is
The individual experience only
Reflection, memory, insight, journey.
Built to be measured
Four signals, read weekly, decide what comes next.
Simplified, not stripped
Each thing I leave out has a reason and a return date.
The hypothesis
A short, personalised reflection loop that remembers you creates an experience people return to, without streaks, scores or reminders doing the work.
Four smaller claims sit inside that sentence. Tap each one to see how V1 reads it.
Tap
01 · Activation
When someone picks an intention, do they actually write?
What I'd read
Share of sessions that end with a submitted reflection. Median length of what they write.
Tap
02 · Resonance
Does the response land?
What I'd read
A one-tap "this landed" or "not quite" under every response. The single most useful piece of data V1 produces.
Tap
03 · Continuity
Does memory make Nudge different from a chatbot?
What I'd read
Resonance on responses that reference an earlier reflection, compared with those that don't.
Tap
04 · Return
Do people come back unprompted?
What I'd read
Day-7 and day-14 return. Reflections per person per week. Across a four-week pilot of 30 to 50 people aged 18 to 28.
How I'd read them
Weekly during the pilot, together. A good V1 is one where all four move in the right direction, not one where a single number looks great.
What I'm not measuring
Streak length, session count, time in app. They reward the wrong behaviour and the brief is explicit that Nudge isn't that.
Every scope decision in the slides ahead points back to one of these four.
The experience · the daily loop
Five screens or fewer. Two minutes a day.
1
Choose
One tap on how you want to show up today. Brave, Clear, Kind, Focused, or simply Myself. No explanation needed, no settings behind it.
2
Reflect
One prompt, not a menu. Written for this person, today, from their intention, their onboarding answers and what they've said before.
3
Respond
A plain text box. No word count, no timer, no "great job" when you submit. Just a place to think.
4
Understand
Four things, always in this shape: an acknowledgement, one insight, a gentle reframe when it fits, one small next action. Then a one-tap "did this land?"
5
Grow
Over weeks, the journey shows the themes Nudge has noticed, grouped by pillar. A story, not a score.
Tuesday · Day 6
How do you want to show up today?
BraveClearKindFocusedMyself
Continue
Clear · Today's nudge
Last week you said the hardest part of the group project was not knowing whether your idea was good enough to say out loud.
Was there a moment today where you had something to say and held it? What was going on just before?
Reflect
Clear · Reflect
In the 11am call, I had a cleaner way to structure the deck but Ritika was already presenting hers and I thought it would look like I was correcting her. So I said "that works" and sent her my version later on chat.
Share with Nudge
Save for later
Clear · Nudge's response
You did say it. Just privately, and after the moment had passed.
That's the third time the thing stopping you has been how it would land for someone else, not whether the idea was good. You already trusted the idea.
One small thing for tomorrow: say the first half out loud in the room. "I had a slightly different structure in mind" is enough. The rest can wait.
VoiceCourage in action
Your journey · 3 weeks
What Nudge has noticed
Holding back when it might land badly for someone else
Came up 4 times · Voice, Courage in action
Clearest when you write before you speak
Came up 2 times · Clarity
The 11am callTue · Clear
Asking for the deadline extensionSun · Brave
This is what Nudge remembers. See everything · Delete anytime
Click a step to see the screen.
The experience · the first two minutes
The feeling I'm designing for: "this asked me something worth answering."
10:00
Sign in
Email or Google. No password to invent. Nothing to verify before they've seen anything.
20:20
Three questions, one screen
What brought you here. Which of five things matters most right now. One thing you'd like to be different. This seeds memory from the very first prompt.
30:50
First intention, first prompt
The prompt already uses their answers. It should feel specific, not like a template with a name dropped in.
42:00
First response
Short. Warm. One insight they didn't expect, one thing to try. Then "this landed?" And they're done for the day.
What I deliberately won't do in session one
No tour. No empty journey screen. No "set up your reminders." No pillars explained. The product explains itself by being good once.
Tone rule
The experience · around day four or five
The moment Nudge becomes Nudge, or stays a chatbot.
A generic AI responds to what you said today. Around the fourth or fifth reflection, Nudge's prompt refers back to something you said earlier, in your own words, and connects it to today.
That's the whole product in one sentence. It's also the thing I'd measure most carefully: resonance on prompts and responses that carry memory, against those that don't.
How memory works, in plain words
Every reflection leaves behind a one-line summary and up to three pillar themes. The next prompt is written from the onboarding answers, the last few summaries, and any theme that has come up twice in the last fortnight. It's a query over your own tables, not a black box. You can open it and read exactly what Nudge knows.
Example · Brave · Day 5
On Monday you wrote that you'd rather send a message than ask in person, because then you don't have to watch the reaction.
Is there something this week you've been meaning to ask someone face to face? What would the first sentence be?
Built from
Onboarding: "I want to stop overthinking before I speak"
Mon reflection summary: prefers messaging over asking in personVoice · Courage in action
Theme seen twice: avoiding live reactionsRecurring
The experience · growth journey and trust
A story, not a dashboard. And nothing hidden about what's remembered.
The growth journey
A timeline of reflections, and above it the themes Nudge has noticed, grouped by pillar. Written as "here's what I've noticed about you," not as charts.
Themes tagged to the five pillars
Every past reflection and response, readable
No scores, grades, streaks or stages in V1
Privacy you can see
The same screen shows exactly what Nudge remembers and why. Trust is part of the experience, not a settings page.
Each person's data walled off at the database level
Nothing identifying is sent to the AI model
Delete my data: one tap, real deletion
Private growth data kept separate from anything a future community layer would touch
The experience · a tiny design system
Calm, warm, unhurried.
Enough of a system to make the first version feel like one product, set up in week one so every screen is built on it. Your brand can replace any of it later.
Colour
Paper
Ink
Sage
Sage tint
Clay
Type
HeadingsA little nudge, for a better you.
BodyWarm sans for everything you read quickly.
RulesRounded everything. One accent per screen. No red, no badges, no numbers bigger than the words.
The eight components the loop needs
Intention chips
BraveClearKind
Prompt card
What's one thing you said today that you'd say differently?
Reflection input
Start anywhere.
Response card + theme tags
You already trusted the idea.
Voice
Resonance tap
Timeline entry · Button
The 11am call
Continue
Scope · what V1 includes
Everything needed to run the loop every day for a month. Nothing that isn't.
1
Sign in and one-screen onboarding
Email or Google. Three questions that seed memory.
Activation
2
Daily intention
Brave, Clear, Kind, Focused, Myself.
Activation
3
Personalised prompt
Written from intention, onboarding and recent history.
Continuity
4
Reflection and AI response
Acknowledge, one insight, reframe, one action. Resonance tap.
Resonance
5
Memory, the lean version
Summaries and pillar themes per reflection; recurring themes feed the next prompt.
Continuity
6
Growth journey
Timeline plus noticed themes by pillar. What Nudge remembers, and delete-my-data.
Return
7
Privacy foundations
Per-person isolation, no identifying data to the model, real deletion.
Trust
8
Pilot measurement
The four signals as saved queries you can run yourself. No admin screens to build.
All four
Where the brief asked for something bigger, I simplified rather than skipped: onboarding is one screen, memory is structured summaries, admin is queries.
Scope · what V1 deliberately leaves out
Each of these has a reason, and a return date.
Private NotebookPhase 1.5
In V1 the reflection is the notebook. You'll know whether people write at all before adding a second place to write.
NotificationsPhase 1.5
The return question is cleaner without them. If people only come back when pinged, that's a different product, and worth knowing.
PaymentsPhase 1.5
Nothing to charge for until the loop is proven. A "founding member" interest list costs nothing and tells you the same thing.
Growth stagesPhase 2
Awareness to Leadership needs real data to define. I tag by pillar now so stages can be derived later.
Android and iOS native appsPhase 2
Earned once retention is shown. Nothing from V1 is thrown away when they come. More on the next slide.
Community, all phasesLater
As the brief says. The data model already keeps private growth data separate so community can be added without a rebuild.
A custom admin panel is also out; the hosted database dashboard and saved queries do the job for a pilot.
Platform · your choice, my recommendation
Mobile-first, two ways to get there.
What you get
A link. Opens on any phone, adds to the home screen, feels like an app. You can put it in front of fifty students next week with no store between you and them.
The honest trade-off
Notifications are weaker: they work on Android, and on iPhone only once the app is added to the home screen. For V1 that's fine, because V1 is testing return without reminders.
My recommendation
Start here. Every rupee goes into the loop, memory and journey. When the pilot shows people come back, the same code goes into the stores as native apps.
Recommended for V1
What you get
Listings on the Play Store and App Store, a proper icon, and push notifications that just work. The familiar shape for a consumer product.
The honest trade-off
Store accounts, build pipelines and review cycles take real effort before a single reflection is written. Within this budget, that effort comes straight out of the product. Apple's review alone can add a week of waiting.
If you choose this
I'd ship the loop and memory, and keep the journey view thinner. It's available as an add-on later for the same price, with no rework, which is why I'd rather earn it.
Possible, not recommended first
At a glance
Installable mobile web app
Android and iOS native apps
How people get it
A link; add to home screen
Play Store and App Store listings
Time to first users
The day the build is ready
Plus store review, typically days to a week
Notifications
Android yes; iPhone after home-screen install
Full push on both
Share of budget on the product itself
All of it
Noticeably less; store setup and review eat into it
Path to the other option later
Same code wraps into native apps, no rewrite
Already there
Use the toggle to compare.
How it's built · the shape, not the wiring
Boring technology, interesting product.
The app
A modern web app that installs to the phone, built on a small design system from day one. Clean, fast, no plugins.
The data
A managed Postgres database with sign-in built in. Each person's rows are walled off at the database level, not just in app code.
The AI
A leading language model behind one small internal service. The model can be swapped without touching the app. Prompts live in your repository, in plain text, so you can read and tune them.
Hosting, your choice
The quick default is a managed setup on a global cloud. The alternative worth a look is E2E Networks, an Indian cloud that tends to cost noticeably less at this scale and keeps your users' reflections on servers in India, a good thing to be able to say about a product built on private data. I can set up either.
Built for Phase 2 without building Phase 2
The web app wraps into Android and iOS apps without a rewrite. Private growth data, notebook and anything community-shaped are separate from the start, as the brief asks.
Running costs, roughly
At pilot scale, the AI model and hosting together come to about ₹100 to ₹300 per active person per month, depending on the model chosen. Billed to your accounts directly, no markup.
Ownership, transparency, documentation
Yours from the first commit. Nothing to hand over, because nothing ever sits with me.
Ownership
All intellectual property is yours. The code lives in a repository under your own account from day one; I work inside it. Hosting, database and AI accounts are opened in your name and I operate them with your access.
100% yours, always
Transparency
You can read every commit as it lands. A demo on a real phone every week. A working build in your own hands from roughly the halfway point, so you're using Nudge before it's finished.
Weekly demo, on a phone
Documentation
I write down technical decisions as I make them: short decision notes, a setup guide and a README. Given the pace and budget, this will be useful rather than exhaustive, and I'd rather say that now.
Useful, not exhaustive
Timeline · about five weeks from kick-off
One dedicated engineer, one product owner (you), one demo a week.
Week 1
Foundations
Repository in your account
Design system and components
Sign in and onboarding
Prompt and tone workshop with you
Week 2
The loop
Choose, Reflect, Respond, Understand
First AI responses in your voice
Resonance tap
Week 3
Memory and journey
Summaries and themes
Prompts that remember
Growth journey
Build in your hands
Week 4
Polish and trust
Privacy screen, delete-my-data
Tuning from your daily use
Pilot queries
Week 5
Pilot starts
30 to 50 people, four weeks
Weekly signal check-ins
Readout and Phase 2 decision
What I need from you before day one
Pillar definitions, a handful of example prompts and responses in the voice you want, tone guidance, and basic brand assets. The memory and response quality is built on this, so there's no slack for it to arrive late.
Your part
Investment
₹3,00,000
Fixed price for the V1 scope on the recommended platform
One number, no hourly surprises. It covers everything on the "what V1 includes" slide, the tiny design system, the weekly demos, pilot measurement, and handover of nothing, because it's all already yours.
Payment terms and start date on the last slide.
Not included, so there are no surprises
Hosting and AI usage
Billed to your accounts, roughly ₹100 to ₹300 per active person per month at pilot scale.
App store fees
Only if Android and iOS native apps are chosen. Google charges once; Apple yearly.
Custom visual design
The tiny design system is included. A full identity, illustration or motion pass is quoted separately.
Anything on the next slide
Each add-on is priced on its own, so you can pick without renegotiating.
Optional add-ons · tap to build your own V1
Daily email nudge
One gentle email a day at a time you set, opt out anytime. Changes what "return" measures.
₹10,000
Notifications on the phone
Android, and iPhone once added to the home screen. Changes what "return" measures.
₹20,000
Private Notebook
Free-form private notes, separate from reflections. Affects activation.
₹30,000
Richer onboarding
Multi-step with a pillar self-check, feeds memory from day one. Affects continuity.
₹20,000
Growth stages
Awareness to Leadership signals from tagged themes, using your rules. Affects the journey.
₹20,000
Payments and subscription
One plan, free trial, Razorpay, cancel flow. No signal in V1; it's a business decision.
₹40,000
Admin panel
User list, usage view, prompt tuning without code changes.
₹30,000
Android and iOS native apps
Both stores, listings, proper push notifications. Plus store fees.
₹60,000
Your V1
₹3,00,000
Base V1, nothing added
Tap any add-on on the left.
A visual design pass (identity, illustration, motion) and community features are quoted separately after a conversation.
After V1 · what the pilot decides
V1 is the first step of a relationship, not the last line of a quote.
A good readout
People write most days they open it, responses land more often than not, memory-carrying prompts land better than plain ones, and a meaningful share is still here on day 14.
Then
Phase 1.5: the add-ons that earned their place, usually nudges and notebook. Phase 2: Android and iOS native apps, early community.
A mixed readout
People write but don't return, or return but say responses don't land. Each pattern points at a specific fix: tone, prompt specificity, or the memory window.
Then
A short tuning phase on the existing build before adding anything. Cheap, because the loop is already there.
A hard readout
Few people write even once. That's a positioning or audience question, not an engineering one, and far better learned for ₹3 lakhs than ₹30.
Then
Everything built stays yours and reusable. I'd help you think about who to put it in front of next.
Next steps
Three things, then we start.
I'd rather spend an hour disagreeing about the hypothesis now than a month building the wrong loop.
1
A call to walk through the hypothesis, the experience and what's left out
Slides 4, 5 and 11. Bring objections.
2
Your inputs: pillars, example prompts, tone, brand basics
A shared doc is fine. Rough is fine.
3
A start date and payment terms
I'd suggest half at kick-off, half when the build is in your hands.