Portfolio demo by Tuan Truong · Hearth & Co. and all data are fictional
HHearth & Co. Opsretail operations demoTry the demo

AI operations for retail chains

Stop calling each outlet to check stock.

Your shops, your warehouse and your online store share one live stock count. When a shelf is about to run out, the system finds the nearest spare stock and asks you to approve the transfer. One tap.

The shop in this demofictional
  • 3outlets: Tampines, Jurong Point, Bugis
  • 1warehouse that also ships online orders
  • 40staff on shifts across 7 days
  • 1,200SKUs of homeware and furniture
0calls to check another outlet’s stock
22:00close report for every outlet, on WhatsApp
S$6.4ksales kept on the shelf in a month (sample)
24/7“is it in stock?” answered by the shop bot

Try it

One stock count for every shelf.

Approve a transfer and watch the numbers move. Then run the weekend and see what runs low next. Everything here is clickable.

H
Hearth & Co. · StockFri 13 Nov · live
Shelves running low4
Sales at risk, 3 daysS$1,341
Stock moved today0 units
Sales keptS$0
Units on hand · days of cover under 3 days fine
ItemTampinesJurong PointBugisWarehouse
Oak bar stool
FUR-112 · S$129
42.0d186.0d99.0d4020d
Queen linen duvet set
BED-204 · S$189
128.0d1010d31.5d258.3d
Rattan floor lamp
LGT-031 · S$249
612d12.0d88.0d1414d
16-piece dinner set
KIT-118 · S$159
99.0d117.3d714d3015d
Bamboo bath mat
BTH-077 · S$29
227.3d307.5d157.5d6012d
Grey wool throw
TEX-090 · S$89
77.0d99.0d21.3d189.0d
Suggested transfers
  • 10 × Oak bar stoolWarehouse → Tampines · 4 left, sells 2/day
    S$1,290
  • 4 × Queen linen duvet setWarehouse → Bugis · 3 left, sells 2/day
    S$756
  • 3 × Rattan floor lampWarehouse → Jurong Point · 1 left, sells 0.5/day
    S$747
  • 4 × Grey wool throwWarehouse → Bugis · 2 left, sells 1.5/day
    S$356
Ask the stock assistant
Activity
  • 10:00 sweep: POS, warehouse and Shopify counts synced. 4 shelves flagged.

Illustrative · fictional sample data

Before / after

Six things that eat a retail week.

The problem and the fix side by side, with what it costs today and what it costs once the system runs it.

Checking stock at another outlet

Today

Call the shop, wait while someone walks to the back room. Customer leaves.

With the system

One live count for every shop, the warehouse and Shopify.

5–10 min a callInstant

Empty shelves on the weekend

Today

Best sellers run out on Saturday afternoon while the warehouse has 40.

With the system

Low shelves are flagged on Friday with a transfer ready to approve.

~S$6k lost / monthMostly caught

End-of-day close

Today

Each manager counts the till, snaps the Z-report and types it into a sheet.

With the system

POS totals reconcile on their own. Only real gaps reach a person.

40 min per outlet~5 min

Owner’s view of the day

Today

Three manager chats and a spreadsheet, read on the MRT the next morning.

With the system

One 22:00 WhatsApp report: sales, gaps, what needs a decision.

Next morningSame night

Rostering

Today

Built on Sunday in a sheet. An MC on Saturday means ten phone calls.

With the system

Shifts follow forecast footfall. Cover requests go out on their own.

3 hrs a week30 min

“Do you have it in stock?”

Today

Instagram and WhatsApp questions answered when someone has time.

With the system

The shop bot checks live stock and holds the item for pickup.

HoursUnder 1 min

Typical figures for a 3 to 5 outlet chain. Real numbers come out of the first week.

WhatsApp

The owner and the shopper on the same stock count.

Grace closes Saturday from her phone at 22:00. An hour later a shopper asks about the same stool, and the bot already knows the van is bringing ten more.

HCHearth & Co. · Ops ●your stock and sales assistant

You are Grace · founder, Hearth & Co..

Press start, then send each of your messages yourself.

What the system did behind the scenes

Nothing yet. Start the conversation and each action appears here as it happens.

AI agents

Three agents. You approve the money.

One moves stock, one closes the tills, one keeps the roster covered. Each stops and asks before anything that costs money or changes a record. Run one and decide.

Stock BalancerWatches days of cover on every shelf and moves spare stock before a best seller runs out

Press Run this agent to watch one of its runs. It stops and asks you before anything goes out.

Does on its own· Recount days of cover per item, per shop· Draft transfers from the warehouse or a slower shop· Book the next van slot once a transfer is approved· Tell the receiving manager what is coming
Asks you first· Any transfer, until you set an auto-approve limit· Pulling stock out of a shop below 7 days of cover· Anything that goes on a supplier order

Rollout

Live in one shop in a week.

Stock first, because that is where the money leaks. Nothing goes live in a shop until the manager has seen it work.

  1. Week 11

    Connect the counts

    Plug in the POS, Shopify and the warehouse sheet, then check the numbers match what is on the shelves in one shop.

  2. Week 2–42

    Stock and the nightly close

    Transfer suggestions and the 22:00 report go live in every shop. Each transfer is approved by hand until the owner trusts it.

  3. Week 5 onward3

    Shoppers and shifts

    The shop bot answers stock questions on WhatsApp and Instagram, and rostering moves off the spreadsheet.

FAQ

What shop owners ask first.

Which POS systems does it work with?+

Anything with a daily export or an API. Shopify, Lazada and Shopee stores connect the same way. If a shop only prints a Z-report, the bot reads the photo.

Will it move stock without asking?+

Not unless the owner says so. Every transfer waits for a tap at first. Later it can move anything under a chosen limit, for example 10 units or S$500.

What if the counts in the system are wrong?+

They usually are at the start. Week one includes a spot count in one shop, and the system flags items whose sales and counts stop making sense, so a shelf gets recounted before it causes a bad transfer.

What does this demo run on?+

This page is a front-end prototype: Next.js, React and TypeScript, with the stock logic, chat replays and agent runs simulated in the browser. A real build connects the same screens to the POS, a database and the WhatsApp Business API.