Get it now

Anxious attachmentearned secure. For the ones doing the work.

Watch yourself become secure.

Vent it, journal it, or paste the spiral. It gets sorted tonight — and week by week, you get an honest answer to "am I actually becoming secure?"

Encrypted in transit and at rest. Never used to train AI models. Never sold, never shown ads. Delete everything with one tap.

They read my message three hours ago.fact They're pulling away, I can feel it.story My chest is tight and I can't sleep.feeling I keep drafting a fourth text.need · reassurance

story ratio 62% feelings honored 1 practice queued

One entry, sorted. Fear separated from fact — in the moment it matters.

Secure attachment isn't something you know. It's something you can do at 1am.

Built specifically for the anxiously attached who are rewiring their brain toward secure — and rewiring isn't done by insight. It's done by reps.

Tonight, reconstructed

You know this night by heart.

Checked if they'd been online.

Reread your last text for the ninth time.

Typed a paragraph. Deleted it. Typed it again.

Pasted the whole spiral into ChatGPT.

"That sounds really painful. Remember — their response time doesn't define your worth…"

Still awake. Soothed for ten minutes. Nothing measured. Nothing trained. Nothing changed.

And honestly? It made sense. Therapy is Tuesday at 3pm. Your friends are asleep. The spiral is tonight.

But comfort that never accumulates is just anesthesia.

The one question nothing will answer: is the work working? This is the instrument that answers.

The instrument

"Am I actually becoming secure?" Now you'll know.

Aware Noticing Practicing Steadying Secure
your road to secure · this weekPracticing
Story ratio, June → now62% → 41%
Recovery after the same trigger3 days → 36 hrs
Direct asks instead of tests+3 this month

Remembering isn't measuring. ChatGPT can recall what you told it — it can't show you where tonight sits on the road from June. This can: months of entries, one scale, one visible arc.

It moves slowly, because real change does. Entries become signals — catches, recoveries, direct asks — and the stage updates weekly. No live number to check at 2am.

Spiraling never lowers it. We measure how you respond to triggers, not whether you got triggered — so you can be honest on your worst night. Nothing resets.

Every move ships with receipts. Not "you're growing 🌱" — numbers, from your own words.

How it works

Four moves. Every day it matters.

New entry · 1:47 am
talktypepaste

screen off works too — just start talking

01Dump it here instead of sending it

Talk it, type it, or paste the spiral you'd have sent to ChatGPT. Sorted in seconds.

Quick rep · 10 seconds
You wrote this on June 3 — "Andrew's pulling away, I can tell"
← storyfact →
Caught it — story ✓"I can tell" is mind-reading.Next time: ask Andrew directly.

3 of 4 caught this week

02Train with cards made from your words

Swipe, get the why and the secure move. Levels climb from fact-vs-story to protest-vs-ask.

Weekly review · Sunday 8pm
Pattern spotted
Andrew online spiral ×9
This week's tool
"Would texting Andrew change anything?"
then hold 10 min

03Spot the pattern, break the pattern

The review names your personal pattern — and hands you the tool that interrupts it.

Your arc · then vs now
Entry · June 8, 1:52 am"Andrew's pulling away. I can't breathe. I checked if he was online all night."recovery: 3 days
Entry · July 20, 11:04 pm"Felt the drop. Held it ten minutes — then gave in and texted. But I came back to center in half an hour."✓ held 10 minrecovery: 32 min

04Watch yourself become secure

Even the gave-in night counts — the app measures the recovery, and your own entries become the receipt.

Where the receipts come from

Every spiral is a story wearing a fact costume. Learn to catch it.

Those numbers aren't vibes. Every entry gets sorted into four honest piles, plus one thing to practice. No endless analysis. No "that sounds so hard."

Fact

What a camera would have seen. "Read at 10:42."

Story

The prediction you're treating as truth. "They're pulling away."

Feeling

Real, valid, never argued with. "My chest is tight."

Need

What the urge is actually asking for. Usually: reassurance.

Why it works: panic needs the story to be true. See the costume and the engine stalls — the feeling stays real, the forecast loses its teeth. You come back to center faster. And it's trainable.

One practice · next 24 hours

"They're pulling away" is a forecast, not a fact. Before any fourth text: say the need out loud once. The ask can wait until morning — and morning-you asks better.

The whole skill stack

Sorting fear from fact is level one. Secure is seven levels up.

Security is what you do — ask directly instead of testing, tolerate not-knowing, talk to yourself kindly. Ten-second reps from your own entries climb there one level at a time.

L1
Fact vs story — catch the prediction wearing a fact costume
day one
L2
Name the distortion — mind-reading, fortune-telling, catastrophizing
L3
Protest vs direct ask — spot the test you almost sent; make the ask instead
L4
The need under the feeling — find what you're really asking for under the panic
L5
Harsh vs kind self-talk — pick the compassionate rewrite
L6
The secure response — scenario in, steadier reply out
L7
Now vs then — is this about tonight, or the old wound?

Levels 3 and up are where the score stops measuring what you know — and starts measuring what you do. Grounded in the published earned-secure research, not vibes.

Chatbots soothe. This one trains.

Leaning on AI for every "am I okay?" outsources the muscle you're here to build. This AI points the other way: it drills you until you catch the story, you name the need, you trust your own read — then it gets out of the way.

Practice makes automatic

Secure isn't never-anxious. It's coming back faster.

Time back to center · after the same trigger
June3 days
Now36 hrs

The reps build a reflex that fires at 1am — without you summoning it. Until it does, just vent:

"he read it and didn't reply, I know I did something, I keep rereading what I sent and…"

Ramble as long as it takes — the mess is the input. Voice works screen-off, saved locally first.

In the light of day

Straight answers, before you give us a single word.

Does AI read my entries?
Yes — that's how the sort works, and that is all it does. Encrypted in transit and at rest, never used to train models, never sold, gone forever the moment you hit delete.
Will it just tell me what I want to hear?
No — that's the loop we're breaking. It won't reassure you about what someone else is thinking, and it won't predict your relationship's future. It shows you your own patterns instead.
How fast will I see something?
Your first sort lands in seconds. Your first Security Stage takes two weeks — the instrument calibrates before it claims anything about you. Real measurement starts slow on purpose; that's how you'll know it's real.
Why not just keep using ChatGPT?
Keep it — for comfort, it's free and it's kind. But memory isn't measurement: no fixed rubric, no versioned score, and an answer that bends toward what you want to hear. This applies one yardstick to every entry for months, and trains the reflex instead of answering for you. "Am I getting better?" needs an instrument, not a conversation.
Isn't this just another AI to get attached to?
That risk is real — reassurance-seeking transfers onto AI easily, and a chatbot that always answers is easy to lean on. So this is built to be hard to lean on: no open-ended chat, one practice instead of endless dialogue, and reps that train your judgment instead of replacing it. You're not meant to trust the app more over time. You're meant to trust yourself more — and open it less.
Is this therapy?
No. No diagnoses, no clinical claims — skill practice and honest measurement, grounded in the earned-security research. If you're working with a therapist, this is the between-sessions gym.
What if my anxiety is right?
Sometimes fear is signal, not distortion. If what you describe sounds unsafe — threat, control, fear for your safety — the exercise stops and says so plainly, and points you to real support.
What happens when I get better?
You graduate, and we celebrate it. Every wellness app is engineered to keep you opening it; ours is engineered so you eventually don't need to. Comfort tapers as your skills grow — for everyone, never behind a paywall.

"I built this during the worst year of my life — journaling spirals into ChatGPT at 2am, asking it if I was getting better. It couldn't tell me. A year of hard work later, I'm steadier. But here's the crazy part: I still can't prove I'm secure. Nobody can. So I'm building the instrument I needed."

— Sid, founder
If you're in crisis or thinking about harming yourself, this isn't the right tool. In the US, call or text 988. International lines: findahelpline.com. Your safety comes before any streak.
unanxious.

You're doing the work.
Watch it work.

Tonight's spiral can be tonight's data. Talk it out — and finally see the line move.

Encrypted. Never trained on. Never sold. Delete everything with one tap.