30+
Beta Users
Tested early builds across iOS and Android
CASE STUDY
AI PRODUCTREACT NATIVE MVPBETA TESTED
01 / OVERVIEW
Mental fitness for
low-energy moments
Flow is an AI-powered mental fitness system for building personalized resets from reusable blocks. Users can combine practices like breathwork, journaling, sound, silence, and AI guidance into Flows that fit the moment they are actually in.
82%
Reported Improvement
In mood, focus, or consistency after using Flow.
4.6★
AVERAGE RATING
From beta users on overall experience
THE PROBLEM
Mental fitness support often fails at the moment people need it most: when energy is low, attention is scattered, and starting feels hard.
MY ROLE
I shaped the product flow, AI experience, prototype, and beta testing around one constraint: the user may not have much energy when they arrive.
THE OUTCOME
A compact AI-guided reset experience that helps users regain calm, clarity, and consistency through small repeatable moments.
02 / PRODUCT IN ACTION
A modular reset system,
built from blocks
Flow is a modular reset system. Users can choose a routine, check in, move through blocks, receive context-aware guidance, and learn whether the ritual actually helped.
01 / CHOOSE
Choose or compose
Start from a saved flow, a prebuilt routine, or build one from blocks.
03 / WHAT WE BUILT
A modular system, not a library of sessions.
Flow was built around reusable blocks that could be stacked into a routine and run as one sequence. The important product decision was making the system flexible without making the user do too much setup.
01
Composable Flow Builder
Users could create routines from blocks instead of relying only on fixed sessions.
02
SIX CORE BLOCKS
Breathwork, meditation, journaling, soundscape, silence, and AI guidance.
03
Prebuilt Flows
Ready-made resets for users who wanted to start immediately.
04
Flow Runner
A shared frame that moved through each block without making users keep deciding.
05
AI Guidance
Generated from journal context only, keeping the feature grounded and restrained.
06
Analytics Loop
Mood lift, consistency, and usage patterns made the ritual visible over time.
04 / PRODUCT DECISIONS
Systematic planning, tough calls, and a viable MVP
We planned the system carefully, but the real product emerged through building. The harder question became what to keep, what to cut, and how to ship something useful before the idea kept expanding.
TRADEOFF01
Less setup, faster starts
The product had to work when motivation was already low, so the path into a flow needed to stay short.
SYSTEM02
Modularity over fixed sessions
Reusable blocks made Flow easier to extend and more personal than a library of one-off routines.
RESTRAINT03
AI only with context
Guidance appeared only after journaling, so the AI had something real to respond to instead of guessing.
MVP04
Ship before it kept expanding
We cut recommendations, wearables, richer journaling, and community features to keep the first version buildable.
05 / WHAT IT TAUGHT ME
The build taught me where product systems become real.
Flow started as a mental fitness app, but the useful lessons came from building the system: how to make flexibility usable, how to keep AI restrained, and how to stop scope from expanding forever.
~/flow/product-lessons.md
READ ONLY
mrinal@portfolio:~$ cd flow
mrinal@portfolio:~/flow$ less product-lessons.md
LOG 01
Flexibility needs defaults
A modular system only worked when users could still start quickly. Prebuilt flows became as important as custom flows.
LOG 02
AI works better with boundaries
The strongest AI moment came after journaling, when the guidance had something real to respond to.
LOG 03
MVPs reveal the real product
Planning helped, but building exposed the real decisions: what to simplify, what to delay, and what to make usable now.
SYSTEM STATUS :OFFLINE