CASE STUDY #03: Companion

Companion: a mental health consumer app

Project: Personal Project

Role: Founder / Product Designer

Team: Only me

Goal: Validate through user testing

Tools: Lovable, Antigravity

Outcomes: prototyped and validated an original, differentiated product

15 - 22 March 2026

Problem: people with mental illnesses struggle to keep track of, and take action on, their personal growth

Someone close to me is affected by this problem. I wanted to to help.

How do target users feel?

Managing treatment is "impossible"
"I don't like thinking about [my personal goals]"
"Ashamed and lazy"

What tools do they currently use?

Finch: self-care virtual pet

iCal: for healthcare appointments

Apple Health: track periods

Clock: alarms for medicine

Insight: Finch doesn't track "progress" and "next actions" though.

New Hypothesis: people with mental illnesses struggle to keep track of the next actions and progress relating to their treatment and goals

Ideation #1

Summary: a cross between a self-care app and a visual novel. You interact with the story through multiple-choice chat.

Sketch #1

Prototype #1

Try it out!

User testing gave me new hypotheses to test:

High Importance:

Users will be more attached to their cat companion if they can create and customise them

Easy to Test:

Visual design is core to this product, because it's differentiated by its emotional impact. Namely, cultivating motivation, encouragement, and an inviting tone.
“Sameen Companion” is a cute name
Users will find narrative more rewarding when it is prompted by the user.

Ideation #2

How might we make the user feel more attached to their cat companion?

How might we help users identify next actions they need to take for each of their goals?

Prototype #2

Try it out!

Findings

The SLM used to generate next actions was not helpful at all.

The main problem was that it failed to recognise nuance. E.g. suggested "do grocery shopping" for the goal of "Skip No Meals". But the actual problem is hunger cues, executive dysfunction and low motivation.

"I would always use it in dark mode if there was one"

The transition from asking the user how they feel to the goals check-in feels "disjointed". E.g, you tell it you feel anxious, and it responds by reminding you of all the things you haven't done.

The visual redesign was a success.

Some goals require habits, not just next actions. The current method for tracking is only suited to next actions, not habits.

Users did not respond well to our simple framework "what is the most obvious next action?"

Redefining the Problem

People with mental illnesses struggle to keep track of the next actions, progress and habits relating to their treatment and goals.
They can find being asked about their progress very distressing.
SLMs are only useful for specific, well-defined tasks.
Our users have a very low tolerance for frustration.