Sumit Dongardive

HOW CLAUDE WAS USED

Kaptiq started as a problem I lived with every day at work. As an APM managing cold storage and HVAC projects, I was constantly stitching together cost tracking, billing, and milestone data across spreadsheets with no single source of truth. I built an Executive Milestone Tracker in Google Sheets first — just to survive — and that became the blueprint. When it came to actually building the product, I used Claude heavily throughout. Not just for code, but for thinking. I’d describe the operational problem — say, how billing timelines rarely match actual cost incurrence on a capital project — and work through the product logic conversationally before writing a single line. It helped me go from “I feel this pain” to “here’s what the data model should look like” much faster than I could alone. On the build side, I used AI to scaffold the frontend, reason through the architecture, and pressure-test feature decisions. A lot of what would’ve taken me weeks of trial and error compressed into days because I could iterate on ideas in conversation first. The core insight Kaptiq is built on — that capital projects need decision-support, not just tracking — that came from my own experience. AI helped me build around it without getting stuck on the parts that weren’t my strength.