Building at the intersection of AI, startups, and product.
I'm Ruchit — a builder who's shipped product across six industries, from AI-native consumer apps to regulated SaaS and emerging-market logistics. I write about what I've learned, share frameworks I actually use, and ship demos for the patterns I claim work.
Frameworks, not hot-takes.
How to evaluate LLM outputs without ground truth
A five-pattern framework for building useful evals when there's no labeled dataset, no obvious right answer, and a deadline.
First ten customers for a B2B AI startup
Why the first ten customers of a B2B AI startup are different from B2B-as-usual — and the four moves that have worked across the AI builds I've seen.
What I learned shipping product across six industries
Six domains — AI consumer, regulated SaaS, last-mile logistics, ed-tech, vertical ERP — and the one through-line that didn't change. A framework you can copy.
Six industries, one through-line: shipping product real users adopt.
Reducing post-capture friction for creators
Creators have plenty of capture tools. The real bottleneck is the post-capture workflow — cutting, scoring, captioning, exporting. AI can collapse a 90-minute edit into 9 minutes if it ships in the right shape.
Personalized discovery for a visual-first category
Helping shoppers find pieces they'd actually buy in a category where taste is the product. Recommendation, shoppable feeds, and AI-powered styling — but in a way the user trusted instead of resented.
Outcomes-driven learning at scale
Most learning platforms optimize for enrollment, not completion or outcome. Designing for the credential employers care about — and for the student who has to finish before they get there — meant rethinking the product.
Operating-tech for dense urban last-mile delivery
Building delivery software for an emerging market — unreliable infra, cash-first economy, operators driving on 2G connectivity. Standard western-stack assumptions broke in week one.
Editorial workflow under regulatory compliance
Pharma content (medical communications, promotional materials) lives under strict regulatory review. The editorial workflow has to ship fast *and* survive an audit — competing constraints most CMS tools punt on.
Workflow software replacing spreadsheets for a specific operator
An industry where operators still ran the business on a stitched-together spreadsheet stack. Generic ERPs didn't fit the workflow; building a vertical one meant going deep on the operator's actual day.
If you're building something at this intersection, I want to hear about it.
Open to advisory work, angel-investor intros, and conversations about AI-native product. The best inbound says specifically what you're working on and what stage you're stuck at.