Case study · 01Soch et Latte · Viman Nagar · Pune

A coffee house's
back-of-house, rebuilt
in six weeks.

Soch et Latte was running a busy Viman Nagar café on a tablet POS, a paper inventory book, and three WhatsApp groups. We replaced all of it with one system the staff actually wanted to use.

Client
Soch et Latte
Industry
Specialty coffee · F&B
Engagement
Fixed-scope build + retainer
Timeline
Six weeks · Q3 2025
app.sochetlatte.in / today
Today
Orders
Prep tickets
Inventory
Customers
Reports
Outlet
Viman Nagar
Today · live orders
Viman Nagar · Pune · live orders
Front of house
Calm
tickets moving
Kitchen
On time
no backlog
Stock
Healthy
auto-reorder armed
#ItemOutletStateTime
#0284Cortado · oatViman NagarPrep0:42
#0283Filter · singleViman NagarReady1:15
#0282Cold brew · 1LViman NagarRouted0:08
#0281Croissant · hamViman NagarPlated2:31
#0280Espresso × 2Viman NagarReady1:48
At a glance

From three WhatsApp groups to one system the staff actually likes.

Order errors are rare nowstaff stopped relying on memory at the bar
Tickets move predictablybar and kitchen see the same queue
End-of-day in minutesstock reconciliation stopped being a midnight ritual
Discovery to launch in six weeksfixed scope, fixed price
01 The problem

Three apps, one notebook, and a lot of yelling across the bar.

Soch et Latte had built a loyal Viman Nagar following in eighteen months. The product was great. The operations were held together with WhatsApp.

Order tickets came in on a tablet POS that didn't talk to the kitchen. Stock was tracked in a paper notebook that lived behind the espresso machine. Supplier follow-ups happened in a WhatsApp group called "supply-chain-final-FINAL." Reconciliation was a midnight ritual the founder did himself, every night, on a spreadsheet that crashed once a week.

The real cost wasn't the wasted hours. It was the order errors at the bar — wrong milk, wrong size, missed customisations — and the silent stock-outs that turned regulars into ex-regulars.

02 What we built

Three pieces that act like one product.

A

Back-of-house order routing

Tickets land at the right station the moment they're rung up. The bar sees coffee. The kitchen sees food. Nobody yells across the counter. Custom modifiers — milks, sizes, syrups, extra shots — render exactly the way they were ordered, with a colour for ready, plated, served.

  • Live order board on iPad
  • Station-aware routing
  • Real-time modifier rendering
B

Inventory by ingredient, not by SKU

Espresso, oat milk, syrups, cups — every item in the menu is decomposed into ingredients with par levels for the outlet. End-of-day stock takes ninety seconds and the system flags low stock before it becomes a stock-out. Supplier orders go out with two taps.

  • Ingredient-level depletion
  • Par levels per ingredient
  • Two-tap supplier orders
C

A customer site that loads

The old marketing site was a Squarespace template with photos that took eight seconds to render. The new site is a fast, editorial Next.js build with the menu, store hours, location maps, and a slim ordering flow that hands off to the back-of-house. Lighthouse 100 across the board.

  • Editorial Next.js front-end
  • Menu pulled live from BOH
  • Lighthouse 100 / 100 / 100 / 100
03 Six weeks, end to end

From a discovery call to a system the staff trusts.

  1. Week 01

    Discovery

    One day on-site at Viman Nagar, observing the bar at peak.

  2. Week 02

    Plan

    One-page spec, fixed scope, fixed price. Signed before any code shipped.

  3. Week 03

    Build · α

    Order routing live in the back-of-house. Real tickets, real prep stations.

  4. Week 04

    Build · β

    Inventory rolled in. Staff trained at Viman Nagar.

  5. Week 05

    Customer site

    Editorial front-end shipped. Old Squarespace decommissioned.

  6. Week 06

    Launch

    Full system live. Retainer signed the same evening.

Stack

Mature, boring, well-supported.

Nothing on the bleeding edge. Everything has documentation, a community, and a future. Soch et Latte's team can hire any competent engineer to maintain it.

Front-end
Next.js · React · Tailwind
Mobile
React Native · Expo
Back-end
FastAPI · Postgres · Redis
Hosting
Vercel · AWS RDS · S3
Realtime
WebSockets · BullMQ
Telemetry
Sentry · PostHog
We told them what was broken. Six weeks later it was fixed. That's still the most surprising thing about working with them.
SLTanmay Gupta · Founder · Soch et LatteViman Nagar · Pune
Up next

More work in progress.

A few projects under wraps until clients are ready to talk publicly. We'll add them here as soon as we can.