SOUTIK!

1. APIs
2. Raw data
3. Summary
4. Dashboard and alerts

Ollvy connects a startup's bank accounts, Razorpay, payroll and analytics, and puts finance, product and payroll on one dashboard.

It flags problems early, like a wrong GST bill, a FEMA fine, transfer-pricing risk between India and UAE accounts, or a sudden jump in vendor spend.

How it works: Ollvy pulls data from each app every night and cleans it up. The numbers come from SQL, not AI. AI just writes the summary on top.

I owned 100% of Ollvy and sold it for $500K.

Built with: Next.js on Vercel, Postgres on Supabase with SQL transforms, edge functions and pg_cron, Claude Sonnet for the narrative layer, and Mistral OCR for documents.

Grew monthly revenue from $30K to $1.5M in 24 months. Acquired 3 companies and grew international markets to the UK and UAE.

Series C · a16z and Peak XV backed

Tavi tells you how your skin is doing using the iPhone's Face ID camera. It builds a 3D model of your face, and it checks 16 things, including acne, wrinkles, dark spots, redness and pores.

Scan again a few weeks later and it shows you what got better. 830 downloads.

Built with: SwiftUI, ARKit, Vision and CoreML

A bill split and expense tracker for groups. Eight currencies with live rates, works offline, and settles up by UPI or card. Receipts, recurring bills, spending stats and CSV or PDF export. 300 downloads.

Built with: React Native and Expo, TypeScript, Supabase, Zustand and NativeWind. The web page is Next.js.

Built and launched a puja booking marketplace: fixed prices with samagri, an auspicious-dates calendar, and SEO pages for every ritual. Puja booking within 1 hour with all paraphernalia. 250 orders fulfilled.

Built with: Next.js and Supabase

Market entry and growth strategy. Payers and providers. Promoted from Associate in 18 months.

Designed the India market-entry model behind MG Motor's launch, which sold about 21,700 vehicles in year one.

Built a health insurer's primary-care growth strategy: 20% lower investment, 40% faster execution, and 4x the expected 10-year EBITDA.

About 10K new paying subscribers a month.

YouTube views from 30M to 150M a month.

Cut $450K a month by renegotiating educator contracts.

About $117K a year saved on an IT procurement budget of about $1.8M.

Built an internal sourcing tool that cut about 12 hours from enterprise approvals.