Personal, properly

She doesn't arrive with a personality. She grows yours.

Most health apps personalize by picking which chart to show you. Milli learns how you want to be spoken to — and then shows you every bit of it, in plain words you can edit or delete.

She asks

A few plain questions up front — no wrong answers, and you can skip the lot.

You tune

Real dials for how she talks, how often, how funny. Change them any day.

She remembers

What helps you gets written down — in the open, in plain words, with her reasoning.

You correct

Wrong about you? Fix it in your own words, or make her forget it entirely.

01 · She asks first

Before she coaches you, she asks.

A handful of plain questions during setup — what kind of support actually helps when health stuff gets messy, whether you want her funny, how deep to go, and any topic she should never joke about. No wrong answers. Skip the whole thing and she'll just learn as you go.

“When health stuff gets messy, what kind of support helps most?”

One of the first things she wants to know about you — before a single number.

02 · You tune her

Real dials, not a vibe setting.

Open “Tune my personality” any day and change how she shows up. These are the actual switches — nothing hidden behind a “conversational tone” slider.

Warmth

Can she call you “friend”? Say she’s proud of you? Gently challenge you?

Humor & emojis

None, light, or playful — and always off in a medication or urgent moment.

How persistent

From a light touch to high support. Never naggy, always easy to ignore.

Direct Mode

A blunter push when you want one — but only after you’ve previewed how it sounds.

No-go zones

Name the topics where she should never joke or push hard. She won’t.

Direct Mode is opt-in twice. She makes you read a sample of how it sounds before it can be switched on — and she quietly pauses it again after a run of hard days.

03 · What she learned

Her memory of you, out in the open.

Everything Milli picks up about you lives in one readable list — plus a weekly “what I learned this week”. This is the part most AI keeps behind glass.

Nothing lands in silence

Every single thing she learns shows up in one list you can read top to bottom. There is no hidden profile.

She shows her reasoning

“Why I think this” sits under each one — the actual pattern that led her there, not a black box.

Sensitive things wait

Anything more personal sits in a “waiting for your okay” tray and shapes nothing at all until you say yes.

Your words win

Correct one and your version is the one she keeps. Or hit forget, and it stops shaping how she shows up.

04 · Nudges that learn

Every nudge can explain itself.

Tap “Why this nudge?” on anything Milli sends and she shows what she looked at — the pattern, the thing you told her, the pace you set. Then four ways to change what happens next.

This was helpful

Keep learning from this timing and style.

Fewer like this

Ease the coaching pace down from here.

Snooze 1 hour

Move it later without changing the category.

Turn off this type

Done with this kind of nudge entirely.

She also learns the shape of your week — where you are, what's on your calendar, when you tend to stall — and backs off on her own when she's getting it wrong.

The part that never adapts

You can change how she sounds. You can't change what keeps you safe.

Personalization sits on top of a safety floor that is identical for every person using Milli. Tuning her tone never reaches it.

The safety check that runs before every answer — it is the same for everyone, every time.

Escalation in an urgent moment. No tone setting softens a “seek care now”.

Medication. She never changes a dose, and humor and Direct Mode switch off entirely.

Sources. However casual you set her, every health answer still shows its work.

More on how that works — Trust & safety →

Then she's yours.

Be first to meet her.

Join early access