Lot 01 — VIBECODE · SHIPPEDof 03

Anumaan

Agentic restaurant operations for independent Indian restaurants — built solo in three days.

Role

Solo — PM, spec and build

Timeline

July 2026 · 3 days

Instruments

Claude CodeNext.jsSupabaseGroqVercel

Live

anumaan-vibeathon.vercel.app ↗

I. Overview

An independent Indian restaurant runs on a WhatsApp group, a paper KOT pad and the owner's memory. Anumaan puts one source of truth across three surfaces — a scan-to-order customer PWA, a kitchen display and an owner's dashboard — and adds an agent layer that drafts a decision, shows its basis, waits for approval, and logs what it did. Built for VibeAthon 6.0, specified as a PRD and FRD first and then vibecoded end to end with Claude Code.

II. Condition Report

The incumbents in this segment digitise the recording of restaurant operations. None of them ship an agent that notices a problem and proposes a fix — the owner still learns paneer ran out when a diner sends back a dish, and still forecasts demand by feel. That gap was the whole product. The hard part was not generating the software; it was deciding what an agent is allowed to do in someone's business, and proving it holds.

III. Product Flow

1.Dine-in order — a diner scans the table's QR code and lands straight on the menu. No app to install, no account. They browse live availability, add to a cart, place the order and watch its status change without refreshing, then settle the bill.

2.Walk-in queue — someone without a table joins the queue from the same code and gets a wait screen, so the host stand stops being a paper list.

3.Kitchen fulfilment — the order appears on the kitchen display as a ticket that staff tap to advance. Agent-created tasks land on the same board and look identical to human ones, so the line learns one surface, not two.

4.86'ing an item — the kitchen toggles a dish off on the availability grid and the customer menu reflects it immediately; the diner sees Sold out rather than ordering something that isn't there.

5.Owner's morning routine — the dashboard opens on a Daily Briefing led by the day's forecast with the figures behind it printed alongside, not a number on its own.

6.Agent proposal and approval — an agent drafts a prep or restock note and shows its basis. The owner approves or rejects it; approved work flows to the kitchen board, rejected work changes nothing, and either outcome is written to the Agent Activity Log.

The scan-to-order customer menu on a phone, showing dishes with prices, an Add to Cart action, and a sold-out dish. — Plate II — scan-to-order menu, with live availability

IV. Appraisal

14/17

eval cases passing — 3 human-graded by design

192

unit tests, plus 143 smoke assertions on the live URL

3 days

solo, from problem statement to deployed product

V. Provenance — the documents this was built from

Written before any code, and readable in full here.

Product Requirements Document

The spec written before any code: problem, personas, scope, requirements, flows, data model and the evaluation strategy.

Read →

Functional Requirements Document

The agent layer and dashboard broken into numbered FR ids, with a traceability matrix, so the build could be graded against the spec rather than a vibe.