You don't have to know what you're asking.
Just say it.

Hold the button, mumble what you're weighing, snap the menu if there is one. WDYT works out the actual question and the options, shows them to you to fix, and only then gives you an answer — with a number for how sure it is.

iPhone · on the App Store
×Take a photo
Corner Diner · Breakfast & Lunch
TODAY'S MENU
Beef noodles$180
Chicken rice$120
Salad$150
Curry pork chop$200
Frame the menu
TODAY'S MENU
Beef noodles$180
Chicken rice$120
Salad$150
Curry pork chop$200
WDYT⚙
TODAY'S MENU
Beef noodles$180
Chicken rice$120
Salad$150
Curry pork chop$200
Listening…
Release to finish
Understanding…
‹Here's what I think you're asking
Q
Which dish on this menu should I have today?
Beef noodles
$180 · rich broth×
Chicken rice
$120 · lean protein×
Salad
$150 · light, high fibre×
Curry pork chop
$200 · fried×
None of these
×
What I picked up
Handwritten menu board, cream background, dark-brown lettering. Four mains: beef noodles $180 (braised broth, comes with a soy egg), chicken rice $120 (Chiayi style, chicken oil over rice), salad $150 (lettuce, cherry tomatoes, boiled egg, vinaigrette), curry pork chop $200 (deep-fried cutlet, Japanese curry). Note bottom-right: free extra rice. You said you're cutting weight and want something light today.
What I know about you
identity.md: cutting weight, no beef, prefers light meals · soul.md: would rather eat less than eat fried food
Confirm and get the answer
Deciding…
‹
Which dish on this menu should I have today?
Salad
✓ confidence 92%
Salad92%
Chicken rice6%
Beef noodles2%
None of these0%
You're cutting weight and prefer light food — the salad fits best.
WDYT⚙
Add photos✓
Up to three
Navy blazer
✓
Beige knit
✓
Black dress
✓
WDYT⚙
Listening…
Release to finish
Understanding…
‹Here's what I think you're asking
Q
Which outfit should I wear to tomorrow's interview?
Navy blazer
$— · formal, sharp×
Beige knit
$— · soft, approachable×
Black dress
$— · elegant, simple×
None of these
×
What I picked up
Three photos on a bed: a navy single-breasted blazer with a white shirt, slim fit; an oversized beige ribbed knit; a black knee-length sheath dress. You said the interview is at a design studio.
What I know about you
identity.md: product designer · soul.md: would rather look understated than overdressed
Confirm and get the answer
Deciding…
‹
Which outfit should I wear to tomorrow's interview?
Navy blazer
✓ confidence 78%
Navy blazer78%
Black dress15%
Beige knit7%
None of these0%
Formal enough for a first meeting, but the shirt-and-blazer keeps it understated — closest to how you like to look.
WDYT⚙
Listening…
Release to finish
Understanding…
‹Here's what I think you're asking
Q
Should I go to the gym today?
What I know about you
identity.md: aiming for three workouts a week, two done so far · soul.md: values consistency over intensity
Confirm and get the answer
Deciding…
‹
Should I go to the gym today?
Leaning yes
? confidence 10%
Not sure — this one is close to a coin flip.
Yes55%
No45%
Probability of yes: 55%. Two of three sessions done; nothing in what you said tips it.
Ask again with this
‹
Should I go to the gym today?
Yes — go today
✓ confidence 76%
Yes88%
No12%
Probability of yes: 88%. Today is the last chance for the third session this week.

Why it feels different

🎤

Speak, don't type

Hold or tap the mic and talk. English through Apple's recogniser; Chinese through SenseVoice running on your phone. No audio ever leaves the device.

✎

It phrases the question for you

A generator model reads your words and photos and writes the question, the options, and what it saw — you never fill in a form.

☑

You stay in charge

Nothing is decided until you've looked. Edit the question, delete an option, add one it missed, correct what it read off the photo. Then confirm.

⚡

A decision, not an essay

The answer comes from TypeSafe's Jev, a model that returns a typed choice with calibrated probabilities in under half a second — no rambling.

½

Honest about uncertainty

When it's 55/45, it says so. When you gave it nothing to go on, it says that too, and lets you add one sentence and ask again.

✦

It learns you, if you let it

Two plain files — identity.md and soul.md — hold who you are and how you decide. Off by default; when on, they're revised after each answer and you can edit them any time.

Three steps, one tap each

Understand → confirm → decide. The decision model is only called after you confirm.

1

Say it (and show it)

“What should I eat…” plus a photo of the menu. Up to three photos; they're resized on the phone before anything is sent.

2

Check what it understood

The question in one sentence, the options with a line each, and “what I saw” from your photos. Fix anything. Nothing has been decided yet.

3

Confirm and get the answer

One pick, the probability of every option, a confidence figure, and one line on why. Add a sentence and ask again if it missed something.

92% — clearPreferences and options line up; take it.
55% — coin flipTold as a coin flip, not dressed up as certainty.
Not enough to go onIt says so and asks for one more sentence.

Your keys. Your phone. No WDYT servers.

WDYT has no backend. Speech is recognised on the device. Understanding goes straight from your phone to Google Gemini or OpenAI with your own API key; the decision goes straight to TypeSafe Jev — via OpenRouter or TypeSafe directly — with your own key. Keys live in the iOS Keychain. Privacy policy · Terms

For fun, not for life decisions.WDYT is entertainment. Its answers come from AI models and can be wrong. For anything that matters — health, money, legal matters, safety, relationships, work — consult a qualified professional and use your own judgment. You use it at your own risk; the app and its developer are not responsible for any decision made with it or any loss arising from one.

Questions

Why two API keys?

Two different jobs. The generator reads your words and photos and writes the question — Google Gemini (key from Google AI Studio) or OpenAI (key from platform.openai.com), your pick in Settings. Jev (TypeSafe) makes the decision — reached through OpenRouter or TypeSafe's own API, so you need a key for whichever you choose. Both are pay-as-you-go on your own account; a typical question costs well under a cent. WDYT never sees either key.

What is Jev, and why not just ask Gemini to decide?

Jev is a “System One” model from TypeSafe: it doesn't generate text, it returns a typed answer — one option from your list, a yes/no probability, or a position on a scale — with calibrated probabilities, in about 100–500 ms. That's what makes the confidence figure meaningful and the answer instant. The generator (Gemini or OpenAI) is used for the part Jev can't do: understanding fuzzy speech and reading photos.

Does the decision model see my photo?

No — Jev is text-only. The generator (Gemini or OpenAI) writes down everything it saw (every menu item with its price, what the outfit looks like) and that text is what Jev judges. You see that text on the confirm screen and can correct it.

Which languages?

The interface is English, 繁體中文 or 简体中文 — your choice, independent of the phone's language. Speech: English via Apple; Mandarin via SenseVoice after a one-time ~230 MB download. The model is shared with other tautiu.dev apps on your phone, so you only download it once.

What are identity.md and soul.md?

identity.md is who you are — diet, budget, habits. soul.md is how you decide — what you optimise for, what you avoid. Both are sent with every question so answers fit you. You can write them yourself, or turn on “update these from my questions” and let the generator revise them after each answer. Off by default.

Is anything stored or shared?

Everything stays on your phone: keys in the Keychain, the two profile files, your history. The only network calls are the two you initiate, from your phone directly to the providers you picked in Settings (Google or OpenAI; OpenRouter or TypeSafe) under their privacy policies.

Can I use OpenAI, and pick the model?

Yes. In Settings → Intent engine, switch the provider to OpenAI and paste an OpenAI key; you then choose a primary and a fallback model — GPT-6 Luna, GPT-6.1 Sol, GPT-5.6 Luna, GPT-5 mini/nano, GPT-4.1 mini or GPT-4o mini — each shown with its price per 1M tokens. The default is GPT-6 Luna, falling back to GPT-5 nano. Grounding with Google Search is Gemini-only.

Stop overthinking lunch.

Download on the App Store