What does Restrologic run for Natural Ice Cream?
Restrologic runs the point-of-sale platform for Natural Ice Cream, an ice cream chain with about 251 outlets trading on it. The platform covers billing, kiosks, a QR menu, Swiggy and Zomato orders, recipe-level stock, reports and an analytics assistant. We also designed a weighing station for the scoop counter.
The challenge
An ice cream outlet is judged at the counter. Queues must move, scoops must weigh what the menu promises, and the owner wants one set of numbers for every outlet, every morning.
At this scale a few things get hard together:
- Volume. The dashboard design measured 251 trading outlets and about 382,000 orders over seven days.
- Growth. The tenant held 285 store records in August 2026, and six had been created in the previous 90 days.
- History. The earlier system covered 4 brands and 302 outlets, with orders from 31 March 2020 to 1 May 2026. That history still had to stay readable.
How did we approach it?
We built the platform as separate services with clear contracts, each owning one job. Orders, catalogue, inventory, reporting, integrations and the gateway are Go services over PostgreSQL. They talk to each other over gRPC.
For history, we designed three temporary line-level reports: order, product and add-on. They run per outlet over a date and time range and arrive as CSV or Excel files. The design is dated 7 June 2026. It was meant to last two to three months and to leave the live reporting code untouched.
What did we build?
At the outlet. The POS terminal handles billing, KOTs, split bills, card payments and day open and close. Captain apps take table orders. An online-order desk receives Swiggy and Zomato orders, with menu sync, item on and off switches, and ratings and complaints from both.
Self-order kiosk. The kiosk takes payment first and creates the order only after the money is captured, so an order exists if and only if payment succeeded. If the kiosk loses its connection, the outcome is still waiting when it returns. The kiosk was first built for one brand. In October 2026 we designed per-brand logo, colours, welcome media and dine-in or takeaway switches, set from admin without a release.
QR menu. A guest scans a code at the outlet and sees a fast, view-only menu on menu.restrologic.com. Sold-out items stay visible and sort last.
Stock and reports. Recipe-level inventory deducts stock from every order. The reporting warehouse backs the dashboards and the report builder.
Analytics assistant. Owners ask questions in plain English and get answers with charts. The assistant reads data only, and its tools are scoped to the asking client’s own tenant and stores. The AI tier went live on 30 August 2026. You can read more about the approach in AI agents for operations.
Scoop counter. A standalone counter screen prints a KOT or bill with a QR code. The ScoopScale weighing station scans that code. Staff then weigh each scoop against its target: 80 g for a single scoop, 160 g for a double, 230 g for a shake and 180 g for a fruit cup. A waffle cone adds 20 g to a scoop. Packs are checked by scanning the tub barcode. See the ScoopScale product page.
What came out of it?
| Measure | Figure | Source date |
|---|---|---|
| Outlets trading on the platform | 251 | August 2026 |
| Orders in a typical week | about 382,000 | August 2026 |
| Swiggy orders in 14 days | 77,162 | September 2026 |
| Zomato orders in 14 days | 100,126 | September 2026 |
| Analytics assistant live | 30 August 2026 | August 2026 |
The Swiggy and Zomato counts are for the business’s main tenant in a 14-day window that ended in September 2026. Across all clients, the platform held about 6.09 million orders over 85 days, from 23 March to 16 June 2026.
The weighing station moved on in September 2026 from prototype hardware to a purpose-built design. It has a custom carrier board that passed its design-rule check with 0 errors, and a 41-line parts list with a supplier code on every line. The firmware port to the new display board was still ahead when this was written.
What is the stack?
The backend is Go, PostgreSQL with TimescaleDB, GraphQL, gRPC and Redis. The admin and kiosk apps are Angular and the QR menu is React. The analytics tier is Python. The weighing station runs on an ESP32.
Timeline
| When | What |
|---|---|
| March 2026 | Platform build starts; order history on the platform begins on 23 March |
| May 2026 | Payment-first kiosk flow designed |
| June 2026 | Temporary reports for the earlier system’s history designed |
| 30 August 2026 | Analytics assistant live |
| September 2026 | QR menu, counter screen and ScoopScale design |
| October 2026 | Kiosk brand settings designed |
If you run a multi-outlet food business, see restaurant technology or talk to us.