01
Export your DMs from Instagram.
Settings → Your activity → Download → request “Messages only” as JSON. We walk you through it in 90 seconds.
Now live
What r
we?
Drop your Instagram DMs. Get a brutally honest spark score, the relationship label you've been seeking, and chat with AI of your crush.
situationship?
delulu?
best friend?
future ex?
soulmate?
almost-something?
ghost incoming?
the one?
situationship?
delulu?
best friend?
future ex?
soulmate?
almost-something?
ghost incoming?
the one?
how it works
Three steps.
One uncomfortable answer.
No screenshots, no copy-paste. We parse your full DM history offline and turn it into a read your group chat would charge you for.
02
Drop the .zip on the upload screen.
Everything is parsed in your browser. We never see the raw messages — only the numbers we need to compute your spark.
03
Read the verdict. Run the chat sim.
Get pacing, initiation %, tone, and a label. Then text “them” with AI to rehearse the conversation you’ve been dodging.
the read
A spark score that
tells the truth.
Every thread gets a 0–100 Spark, a one-line verdict, and the receipts to back it up. Here’s a real-looking sample.
kendalljenner · 111 msgs
warm but
unclear
● Mid-spark · ambiguous intent
kendalljenner shows engaged interest through frequent initiation (60% of conversation starts), quick replies (25 min avg), and genuine enthusiasm about meeting up. However, the signals remain ambiguous — there are zero direct romantic or affectionate words, the tone is playful-friendly rather than flirtatious, and kendalljenner deflects with humor.
60%
Initiated
25m
Avg reply
0
“♥” words
what we label threads
From soulmate
to soft-ghost.
Eight verdict categories trained on real conversation patterns — pacing, vocabulary, initiation balance, and reply velocity.
Mutual spark85–100
Slow-burn crush70–84
Warm but unclear50–69
Friendzone, comfy35–49
Soft-ghost incoming0–34
chat simulator
Chat with your
AI crush.
We turn each DM thread into an AI persona that texts the way they text — pacing, slang, emoji habits, deflection style. Practice asking the awkward question before you ask it for real.
-
Their voice, not a chatbot.Trained per-thread on actual phrasing, reply gaps, and tone.
-
Try the hard question.“Is this a date?” “Are we exclusive?” “Do you still like me?”
-
Switch personas instantly.Toggle to a “Playful” or “Brutally honest” alias of the same person.
T
kendalljenner · AI
Playful persona
Oh~~when r u in LA??
maybe next month 👀
tell me first ok?? we should grab dinner
privacy first
Your DMs
never leave
your device.
This is the part that matters. Reading romantic signals from someone’s messages is intimate. We treat it that way.
Parsed in your browser
Your .zip is unpacked locally. Raw message text never touches our servers — we only ever see the aggregate numbers needed to compute your spark.
One-tap delete
Every thread has a Delete button on its insight card. Hit it and the analysis is wiped from your account in the same second.
No training, ever
Your messages are not used to train anything. The chat sim runs against a per-session embedding that’s deleted when you close the tab.
No tracking, no resale
No analytics on your reads. No data partners. We make money when you top up — that’s the whole business model.
ENCRYPTION_AT_REST
NO_RAW_TEXT_UPLOADED
SESSION_ONLY_EMBEDDINGS
NO_RAW_TEXT_UPLOADED
SESSION_ONLY_EMBEDDINGS
We can’t read your DMs even if we wanted to.
pricing
Pay only when
you want to know.
Top up a tiny credit balance. Each analysis and chat run costs cents. No subscription. No surprise bills.
Free taste
$0/forever
Get one free Spark read on the house.
- 1 thread analysis
- Spark score + verdict
- Privacy-first parsing
- No card required
$5, $10, $20 top-ups available. Payments powered by PayPal
faq
Most Asked Questions
Is this actually private?
Yes. The .zip is parsed in your browser using JS — raw message text never goes to our servers. Only aggregate numbers (counts, average reply gap, vocabulary buckets) are sent for the spark calculation. Nothing is used for training. Hit Delete and it’s gone.
How accurate is the spark score?
Accurate enough to be uncomfortable. We score initiation balance, reply latency, message length symmetry, vocabulary affect, and pacing changes over time. It’s not a love oracle — it’s a mirror for patterns you’ve already half-noticed.
Can I analyze a thread that includes someone else’s messages?
Technically yes, since their texts are in your DM export. We strongly recommend only running analyses on people you’d be comfortable telling. Don’t be weird.
What does the chat simulator actually do?
It builds a per-thread persona — vocabulary, tempo, slang, emoji habits — and lets you message it freely. Practice the hard question, see how “they” might dodge, then go say it for real.
Will this tell me if I should text them back?
No. It will tell you what your DMs already say about the situation. The texting back part is on you.
What does it cost?
One free read to start. After that, top up $5/$10/$20 in credit. Each analysis is ~$0.10. Each chat-sim message is ~$0.03.
1 free read · no card
So… what r you?
Find out in two minutes. Drop your DMs, get the verdict, run the sim. The group chat will be jealous.