From a 12-hour hackathon MVP to an app 50+ reps use

01

The problem

After every site visit, a Truva sales rep had to fill in a long form in the Zoho CRM mobile app to record what the buyer was looking for.

Only 35% of visits were logged the same day. For the rest, the buyer's preferences were lost. Without those preferences, Truva could not match buyers to the right homes.

02

Who the users were and what I learned

I shadowed up to 10 sales reps on their visits. The form was the problem, not the reps. It was long, and filling it in on a phone after every visit was easy to put off.

Two things had to change together: logging had to take less effort, and reps needed a nudge to do it.

03

Options considered

1. A shorter form. Less effort for reps, but less data about each buyer.
2. Chosen
   Voice notes turned into the full form (chosen). The rep speaks a note, and an LLM fills in the structured form. The form stays complete, but the rep no longer has to type it.
3. What to leave out. To ship in 2 months instead of 3, I dropped two features that did not matter: automatic highlighting of summaries, and recording that runs all the time, even in the background.

04

What shipped

The hackathon version. In February 2026, Truva ran a company-wide AI hackathon with 25+ teams from engineering and business. I led a team of 5, and we built the first version in 12 hours. It won, with a ₹1.5L prize.

The production app. I wanted the hackathon app live before I left Truva. It was my baby, so I built the production version myself in React Native. It has:

• Voice capture: the rep records a note, and the LLM fills in the visit form in the CRM.
• Automated reminders, so a visit does not go unlogged.
• A daily visit calendar, with customer details and talking points for each buyer.

The launch also mattered beyond the app. I was running the tech team at the time, while the CTO raised the Series A. Shipping this in 2 months instead of 3 is one of the reasons the team beat its delivery targets by 30%.

05

Result

• Same-day logging rose from 35% to 80%+.
• 50+ reps use the app.
• Visit-to-deal conversion rose 10-15%, adding ~₹2Cr of listings.

That last result is a chain, so here it is step by step. Easier logging meant 80%+ of visits were logged the same day. That gave Truva better data on what each buyer wanted. Better data meant better matching of buyers to homes. And better matching meant more visits turned into deals.

06

What I'd do differently

I would bring a second engineer onto the production app early. Building it alone got it shipped before I left, but a co-owner would have made the handover easier.