Feature I  ·  Talk to log

Most apps ask you to navigate a form during your worst moment.
We ask you to whisper.

Natural-language logging is the wedge. Vaely takes a sentence you would say out loud — "took 400mg ibuprofen at six pm, head pain seven, nausea coming in" — and turns it into a structured journal entry. The on-device AI does the structuring. You confirm what it heard before it saves. Anything it got wrong, you fix in one tap. Nothing is logged that you did not see.

A side-by-side

Twelve buttons, or one sentence.

A. A leading symptom tracker. To log one event:
  1. Open app
  2. Tap Log
  3. Tap Symptoms
  4. Scroll the symptom list
  5. Select Headache
  6. Drag severity slider to 7
  7. Tap back
  8. Tap Medications
  9. Search for Ibuprofen
  10. Select dose
  11. Set time
  12. Tap Save

Twelve taps, give or take. During brain fog. While in pain.

B. Vaely. To log the same event:
"Took 400mg ibuprofen at six pm, head pain seven, nausea coming in."
What Vaely extracts
  • Time6:00 pm
  • MedicationIbuprofen 400mg
  • SymptomHeadache · 7 / 10
  • SymptomNausea · onset
Tap to confirm and save.

One whisper. Fifteen seconds. Your hands free.

I'm too tired and sick to press a hundred buttons on my symptom tracker app. — Patient verbatim, public chronic-illness communities

We did not invent that frustration. We listened to it. Voice-first logging is the response.

Four ways in

However you can speak today.

i

Voice on iPhone

Tap the microphone, speak the entry, watch the structured fields fill as you talk. Apple's on-device dictation handles the words; the on-device LLM handles the structure.

ii

Voice on Apple Watch

Raise your wrist. Tap. Speak. The watch app runs the same logging flow as the phone, with the structured entry appearing on both devices when you confirm.

iii

Typed entry

Some patients prefer to type. The same on-device LLM extracts from a typed sentence. Same confirmation step. Same structured output.

iv

Siri shortcut

"Hey Siri, log my headache." The basic shortcut is in the free tier. Advanced shortcuts in Pro accept complex structured speech: "Log migraine severity seven, took ibuprofen four hundred milligrams, started two pm."

The trust step

Nothing is logged that you did not see.

Every voice or text entry produces a confirmation card. Each extracted field is shown — the medication, the dose, the time, the severity, the symptoms. You scan it. You correct anything that drifted. You tap save.

The AI is not the judge. You are. The confirmation step is the trust step, and it is non-skippable in a way that respects the patient instead of the developer.

What gets structured

A taxonomy in your own words.

Time of event
"At six pm" / "this morning" / "around lunch" — relative or absolute, the AI resolves it.
Medication, dose, route
"Ibuprofen four hundred" / "two more naratriptan" / "the same as yesterday."
Symptoms (one or several)
"Head pain" / "joints stiff" / "fatigue is bad today" — multi-symptom in one sentence is fine.
Severity (1–10 scale)
"Pain seven" / "a bad five" / "manageable" — qualitative phrases map to the standard scale, with the original wording preserved.
Side effects
"Made me dizzy" / "nausea coming in" — coupled to the medication when context is clear.
Mood and energy
"Drained" / "fine actually" — captured when mentioned, never asked for.
Free-text notes
Anything the structured fields miss. Your phrase, kept verbatim, searchable later.

Multi-language

Speak in the language that fits your day.

The on-device language model handles many languages for natural-language extraction. English, Spanish, Turkish, German, French, Portuguese, and others. The website is English at launch; the app is multi-language. A Spanish-speaking patient in El Paso and an English-speaking patient in Manchester get the same extraction quality.

Accessibility

Voice-first is the accessibility story.

Before VoiceOver, before Dynamic Type, before Switch Control and AssistiveTouch — the fundamental accessibility move is a voice-first input paradigm. A patient with brain fog, with hand tremor, with photophobia, with a flare in both wrists can log a full entry without navigating an interface. Vaely supports the standard iOS accessibility suite, and chooses voice-first as the load-bearing decision.

Questions about logging

The plain answers.

What if the AI mishears me?

Every entry shows you exactly what was extracted before it saves. You can edit any field, remove anything that does not belong, or scrap the entry and try again. Nothing is logged that you did not see. The confirmation step is the trust step.

Does it need an internet connection to work?

No. The on-device language model handles dictation and structuring locally. The app works on a phone in airplane mode, on a plane, in a basement clinic with no signal, in a car — anywhere your phone has battery.

How long does logging take?

About fifteen seconds for a typical entry. You speak in your normal cadence, the structured fields fill in as you talk, and the confirmation card appears when you stop. Compare a half-minute conversation with your phone to the alternative — twelve taps through nested forms during a flare.

Can I still log by tapping if I prefer?

Yes. Manual structured entry is available for everyone, every time. Some patients prefer it for routine medication logs. Some prefer it on calm days. Voice is the wedge feature, not the only feature.

What languages does it understand?

The on-device language model supports many languages for natural-language extraction. Speak in the language that fits your day. The website is English at launch; the app is multi-language.

What does it actually extract from my words?

Time of event, medication name, dose, route, symptoms (one or several), severity on a one-to-ten scale where you mention one, side effects, mood and energy if you mentioned them, and the rest as free-text notes. The structured fields make patterns findable later. The free text preserves your voice.

Whisper a sentence.
Vaely does the rest.

The free tier includes unlimited natural-language logging. We do not gate the feature that turns "I forgot" into a record.

Download on the App Store