Voice input pilot · production proposal

No data entry.
They just speak.

You said it plainly: everyone hates data entry, and the version of your app that wins has none at all. The customers who already stick are the ones speaking their day into WhatsApp. This turns that into the whole product. Open it, speak, done. It runs in production and feeds clean, patient-tagged data straight into your system.

Prepared for the September 10 call · HumLife360

9:41
HumLife360
TODAY Not logged yet
HumLife360
Hold to speak
Say "sugar was 140 before lunch, walked 20 minutes"
1 tap
to log a whole day by voice
0 forms
no fields, no menus, no navigation
3 of 4
customers Indian, handled in their own accent
~50
customers live in the production pilot

01 · The opening

The friction you named is the whole opportunity.

The next version of our app will not have any data entry feature at all. The only thing you can do is going to be voice.
Harsh Mulik · September 9 call

Data entry is the step every customer drops. The ones who stay are already speaking their day into WhatsApp, and you piece it together by hand at about 85 to 90 percent.

The pilot turns that habit into a real product. One tap, they talk, and the day arrives structured and attributed on your side. No forms, no menus, nothing to navigate.

02 · The experience

The whole thing is three taps.

This is the app we would build for you. Every screen here is real to the pilot.

9:41
HUMLIFE360
hold to speak your day
Opening HumLife360…
LISTENING
Fasting sugar was one forty this morning. Had two rotis and dal for lunch. Felt low on energy in the afternoon.
Here is what I heard
Fasting glucose140 mg/dL
Lunch2 rotis, dal
EnergyLow, afternoon
Looks right
The interaction One tap, and the day is logged

This is the customer capturing their day in one tap.

9:41
Today
Yesterday 8:12 pm · Fasting 132, oats, energy good
Today 9:41 am · Fasting 140142, 2 rotis, dal, low energy Saved
Inside the app Every entry stays reviewable

Everything captured is reviewable and editable inside the app.

03 · How it works

Words in one end. Clean records out the other.

Structured, patient-tagged data comes out, and your developer team pulls it into your existing app on their own schedule. Nothing we build touches what you already run.

Speak
Voice note
One tap, in their language.
→
Transcribe
Accent-ready
Indian accents and English, handled right.
→
Structure
Named fields
Glucose, meals, symptoms, mood.
→
Confirm
Review + tag
Checked, tied to the patient.
→
Your side
API → your team
They ingest into your database.

04 · Accuracy

The problem that broke the last attempt, designed out.

A previous vendor lost records because their engine could not handle South Indian names, and the database kept throwing back unknown user. Two decisions remove that failure.

How it broke before
Speak yourname Western speechengine Namegarbled Unknownuser
Record dropped
How we do it
Log inonce Speak yourday Accent-readyengine Tagged to theright patient Saved
Record saved

05 · The technology

What we use, and why it is the right tool.

Every choice here earns its place. No technology for its own sake.

Indian accents

Accent-ready speech

It gets Indian names and numbers right. We add a speech engine (Sarvam) tuned for Indian accents on top of our English pipeline, so a name a Western engine would garble comes through clean. For a base that is three quarters Indian, that is the difference between data that lands and data that gets thrown out.

Already proven

Our English engine

The same voice pipeline we run in production today. For English-first customers we use the stack that already powers Mooney, our own live app, so there is no guesswork on the part we have shipped before.

Speech to fields

Extraction layer

Free speech becomes clean, named fields. A rambling account of a day turns into glucose reading, meals, symptoms, and mood, ready to store.

You pull, we do not push

Your API

We expose the data, your side ingests it. Your developer team reads clean records from a simple interface and loads them into your database when they want. It sits alongside your app without changing anything inside it.

06 · Proof

We already run this.

Mooney is our own product, live on real phones. People speak their day and it files a clean, structured entry. The voice engine under your pilot is the same one running here.

Mooney home screen

The Mooney home screen. Every entry here was spoken, never typed.

See it in motion. A short walkthrough of speaking an entry and watching it file itself.

Try it yourself at usemooney.app

07 · What we need from you

Three small inputs, and the accuracy takes care of itself.

Input 1

A one-time login per customer

So every voice note is certain to belong to the right patient. They sign in once, then just speak from then on.

Input 2

The attributes to capture

Which fields we pull from speech for the pilot, a core set to start, growing toward your full set of attributes over time.

Input 3

The handoff to your developer team

A short agreement on the interface their side reads from, so records flow into your database with no change to your app.

08 · The plan

One production build, in two stages.

Start small and real, prove it on a handful of customers, then scale on what the pilot actually shows. Every stage ships something live.

Stage 1

Production pilot

Live on about 50 customers
  • One-tap voice app with a home-screen widget
  • English plus Indian-accent handling, tuned for names and numbers
  • Core attributes: glucose, meals, symptoms, mood
  • Review-and-edit before save, patient-tagged
  • API handoff so your developer team ingests automatically
Stage 2

Scale and deepen

Scoped from the pilot results
  • Full regional-language range across your base
  • More attributes, toward your full set
  • Second platform for the rest of your customers
  • Hardening and monitoring at real volume
  • Committed only once the pilot has proven itself

09 · Pricing

You see exactly what you are paying for.

Stage 1, line by line. No lump sum, no mystery.

Voice capture app and one-tap home-screen widgetThe Mooney mechanism, built for your flow.
$1,200
Voice integration and accuracy tuningSpeech tuned for Indian accents, names, and numbers, on our proven English engine.
$1,100
Extraction engine: speech into structured fieldsLoose speech becomes named health records.
$900
Review, edit, and patient taggingAccuracy checked at the source, tied to the right person.
$600
Data API and handoff to your developer teamClean records their side pulls automatically.
$400
50-customer pilot setup, testing, go-liveStanding it up on real customers.
$200
Stage 1
$4,400 to $5,000

$4,400 covers the 50-customer pilot. A larger group scales toward $5,000. These are starting numbers, and we are open to shaping the scope and the price with you.

Includedfor the pilot

Running the pilot is part of Stage 1. Speech transcription, hosting, and accuracy tuning through the pilot are covered, so there is no separate monthly fee while you are trying it out.

Stage 2, scaling, is scoped from the pilot in the range of $4,000 to $6,000.

New features beyond this scope are priced on their own, and you always see the breakdown first. Company registration is on our side to complete before we begin, which is in progress.

Back to overview