Projects / Family business

Uni-Nation Digital.

Being repaireduni-nation.in / a deploy shipped without a required build setting and locked every user out. The fix is a guard so it cannot ship that way again

My family has manufactured uniform fabrics in Maharashtra since 1988. I built its digital side: a public catalogue, an inventory system, and stock entry that works the way the warehouse actually works: by voice, in Hindi.

1988established
~2,000products tracked
650+customers
~$2mof sales data managed
The problemSince 1988

P. Ashok Kumar is the business, named after my grandfather: uniform fabrics for schools, companies and industrial workwear, made in Ulhasnagar since 1988 and sold from an office in Kalbadevi, Mumbai. My dad built it over three decades. I grew up around the looms and the buyers who wanted the same shade of navy they got last year. After college I joined as Accounting and Sales Manager and pushed us into finished garments, then moved to the UK for my master's.

For decades it ran on paper and trade relationships. No website, so new customers couldn't find us, and a stock ledger kept in notebooks. I still help with the bigger finance and strategy calls, and I built and run everything digital.

The catalogueuni-nation.in

The business ran on word-of-mouth and trade relationships for decades. No website, no digital catalogue, no way for new customers to find us. uni-nation.in changed that: fabric detail pages, colour swatch selectors, downloadable shade cards, and enough SEO to actually show up when someone searches for uniform fabric manufacturers in India.

  • Digital swatch selectorDistributors used to rely on physical swatch books shipped across the country. The digital catalogue filters hundreds of variants by composition, GSM and weave, instantly.
  • Static and fastNext.js static generation keeps the catalogue under 400ms on India's wildly uneven mobile networks. Each product page renders its own weave pattern in CSS: knit loops for Matty, diagonal twill for Samray.
  • B2B lead pipelineThe sample-request flow routes distributor leads straight to the sales team's inbox and WhatsApp. Lead capture through Formspree, eighteen shade card PDFs behind it.
Uni-Nation catalogue site on desktop
uni-nation.inDesktop
Uni-Nation catalogue site on a phone
Mobile
The inventory systemVoice first, Hindi first

The stock ledger lived in notebooks and in people's heads. Around 2,000 products across four fabric lines and their colourways, and the only person who knew what was actually on the shelf was whoever last walked past it. Standard inventory software failed here for one physical reason: warehouse workers hold fabric rolls in both hands. Nobody is stopping to type into a form.

So voice is the input. A worker taps once, says what came in or went out, in Hindi, and the system does the rest. The parser understands fabric units like thaan, taaka and metre, reads Hindi numbers, and corrects the speech-to-text errors that fabric names always cause. Behind it sits a 30,000+ line PWA with the sales history cut into dashboards: which lines move, which customers slow-pay, what to weave next. About $2m of sales data, tracked properly for the first time.

Voice stock entrySample exchange

"Samray navy blue, teen thaan aaya"  (three rolls of Samray navy came in)

PARSED  product: SAMRAY / NAVY  |  qty: +3 thaan  |  type: inward  |  STOCK UPDATED
Play a stock entry spoken in Hindi.
The productFour fabric lines
LineConstructionUsed forSpec
MattyPique-knit, 100% polyesterSchool polos, sportswear, event tees. The workhorse180 GSM / 58"
SamrayTwill, poly-rayon 65/35School trousers, blazers, formal sets. The best-seller, nineteen colours220 GSM / 58"
TrovinePlain weave, poly-viscose 70/30Hospital scrubs, industrial workwear, government contracts200 GSM / 60"
FilafilEnd-on-end, poly-cotton 80/20Dress shirts and office wear. Looks more expensive than it is160 GSM / 58"
Four fabric linesDistribution across India
Stack
Next.js 16Catalogue site: App Router, server components
Tailwind v4Styling on both surfaces
GSAPCatalogue scroll animations
WhisperHindi speech-to-text
PythonVoice parser: units, numbers, corrections
PostgreSQLStock ledger and sales history
Node.jsPWA backend
VercelCatalogue hosting and deploys
Google CloudPWA and voice infrastructure
Other projects
RefProjectSummaryStatus
01 OpsPilot The AI that answers my messages: 6,124+ documents, live finance data, one-click approval. Production
02 Trading Bot Four AI agents that veto my trades. 115+ sessions, zero impulsive trades let through. Paper trading
04 WhatsApp Smart A Chrome extension that gives WhatsApp scheduling, AI replies and summaries. Prototype
05 SCS Cockpit The study engine I built for my own CIMA Strategic Case Study. I passed. Retired, job done