Concept demo
Jar × Dvar · concept demo

More savers saving, every single day

AI that writes every Jar user a message about their own streak, goal and reward, keeps the habit alive when they try to cancel, and explains their payout before they worry. Built to lift conversion and retention, in all 8 of Jar's languages, on models Jar owns.

Start the live demoWhat it is worth
35M+
savers, most of them saving for the first time
They need a reason to save today, not the same template as everyone else. Registered users, company figure.
1M+
autopays a day, each one a habit to keep
Every cancel turned into a smaller step keeps recurring savings flowing. Company figure from a public job post.
8
app languages, one personal message in each
Personal at this scale is not possible with hand-written copy. It is with AI.
Three levers, live

Pick one and run it with Jar in the room

ConversionA personal message for every saver

Push, WhatsApp and SMS written from each user's streak, goal and reward, in their language. More users who save today, and lapsed users who come back.

Save rate · reactivation 8 languagesWrite messages RetentionKeep the habit when a user cancels

Shows what they have built and offers a smaller step that fits their reason, instead of losing the autopay. Cancel stays one tap away.

Autopays keptRecurring savingsTry a cancel Trust“Why is my payout lower?”

The top complaint in Jar's reviews, answered with the user's own GST, spread and price change before they withdraw in frustration.

Ratings · fewer ticketsUsers who stayAsk My Gold
What it is worth

A few more saves a day, across Jar's base

Move the sliders to Jar's numbers. The first A/B test replaces every assumption here with a measured one. The save flow has the same view for autopays kept.

Jar has 35M+ registered users
A personal message vs today's template; measured with an A/B test
Jar savings start at ₹10
Extra saves per day5,000
Extra savings per month₹75.0 L
Per year₹9.00 Cr
Even at half the lift₹4.50 Cr a year

Every extra save is gold bought through Jar, the flow its revenue comes from. Counts only the day of the message, not the habit it restarts.

Why Jar will need this

The apps that win the next 35M savers will talk to each one personally.

Jar's latest release already moves the home feed towards personal updates. Doing that in every message, in every language, needs AI, and the earlier Jar starts, the more its own data teaches the model.

1
Templates have hit their limit35M users, every lifecycle stage, 8 languages: hand-written copy cannot follow each user's streak, goal and reward.
2
Trust drives repeat savingExplaining charges in the user's own numbers protects ratings and keeps people saving after their first withdrawal.
3
Safe by designApproved claims only, Jar's numbers only, every message checked before it is sent.
4
Costs stay under controlOpen models Jar owns, run on Dvar in India, already 7.8× below a frontier API per message, and lower again once trained for Jar.
How Jar gets there

Prove the lift on one flow, then scale it

The full plan
1Baseline

Pick the metric (save rate on messaged users) and capture today's messages with Dvar Eval Bench.

Week 1
2A/B test personal messages

A model trained on Jar's campaigns, tone and claims, against today's templates.

Weeks 2 to 6
3Scale what wins

Run it on Dvar in India, with checks on every message and a predictable cost.

On the measured lift
4Expand

Save flow and My Gold next, then SMS round-ups and support.

Next quarter