The platform

One platform that sells, operates, and remembers.

Not a feature bolted onto your store — a system. Up front it talks to shoppers, in back it watches your data, and in the middle you stay in charge.

  1. 01Customer AIA shopper asks for a waterproof jacket. Nothing in the catalog matches, and the assistant says so.
  2. 02Multi-model engineA product question — routed to the fast model.
  3. 03Agent MemoryThe ask is kept as a conversation signal.
  4. 04Ecommerce AgentsDemand Radar, 07:00: six asks this week. Unmet demand.
  5. 05LumnyaLumnya brings it to you, with the conversations as evidence. You decide.

Two journeys through the platform on a demo store. Pick one to follow it — figures are illustrative.

Up front, the Customer AI talks to shoppers. In back, the Ecommerce Agents scan your data and propose actions. Lumnya sits in the middle, where you ask and approve. Underneath, Agent Memory stores what's learned — and a multi-model engine routes each task.

Four pillars

The system, piece by piece.

The storefront assistant that sells and supports — 24/7, multilingual, grounded in your policies.

Ask in plain language, get answers from live data, and approve the actions that matter.

Specialist agents that scan your store on a schedule and propose what to do next.

Private notebooks, shared verified facts, and a fresh snapshot — intelligence that compounds.

aurora-supply.com · storefront chat

Do you ship to Cyprus?

Yes — 3 to 5 working days, tracked.

from your shipping policy

Έχετε το hoodie σε M;

Ναι, το Nordic Hoodie σε M είναι διαθέσιμο. Να το προσθέσω στο καλάθι σας;

answers in the shopper’s language

24/7 · no ticket opened · nothing invented

Lumnya · your manager

›

What will change

Oxygen Pro (EUR)520.00→572.00

Yes, apply itwaits for you

Done — journalled, and one click undoes it.

this morning’s scans

  • Inventory Forecasterdaily 06:00

    scheduled

  • Profit Lensdaily 07:00

    scheduled

  • Demand Radardaily 07:00

    scheduled

  • Liquidation Plannerdaily 08:00

    scheduled

each finding opens a recommendation, with the reasoning attached

Private notebook · Inventory Forecasteronly this agent

Hoodie / M sells ~3 a day since the restock. Lead time 9 days.

Shared verified factsevery agent reads

  • Oxygen Pro — keeps 38% after costs
  • Nordic Hoodie / M — stock critical

Fresh snapshotrefreshing…

Demo store; figures are illustrative.

The engine

The right brain for every task.

Fast models for routine work; stronger models for revenue-critical moments — checkout, cancellations, returns; the most capable for high-stakes reasoning. Speed where it's cheap, depth where it matters.

multi-model engine

routine work → fast modelsrevenue-critical moments → stronger modelshigh-stakes reasoning → frontier models

router · reads the intent

Fast model

routine work

    Stronger model

    checkout · cancellation · returns

      Send one

      The operating loop

      Recommend. Approve. Act. Remember.

      The rhythm of running a store on Lumnya — and each cycle makes the next recommendation better.

      Built on trust

      Powerful, and always in your control.

      Per-store isolation — your data is never pooled or shared.
      Confirm-gated actions on anything touching money or a customer.
      A full audit trail — every action logged with its reasoning.
      Agents default to recommend; autonomy is granted by you.

      Who it’s for

      Built for growing stores, not 40-person ops teams.

      SMB and mid-market Shopify merchants who need leverage, not another dashboard to babysit.

      Private beta

      Explore the platform piece by piece.

      Start where your customers are — the front of house.