Ecommerce agents, explained
An ecommerce agent is AI software that works inside an online store. It reads the store's own data, decides what should happen next and does it: answers a shopper, flags a product about to sell out, proposes a discount on stock that stopped moving. The trustworthy kind asks before anything touches money or a customer, and logs every action.
Liquidation Planner · daily 08:00
scanning01 · Observe
Reads a fresh snapshot of the store
Catalog, stock and orders, from a private snapshot of your data. It never hammers Shopify to get them.
Snapshot · read-only
One loop of a back-office agent, the way Lumnya's Liquidation Planner runs it. Tap a station to hold it.
Software has spent twenty years telling store owners what was wrong. An ecommerce agent is the first kind that does something about it, and shows its work.
The definition, unpacked
Plenty of software has one of these. An agent is the thing that has all three, and the trust controls that make the third one safe.
Three kinds
Most talk of ecommerce agents means one of three things. Two you can put to work; the third is already visiting.
01 · Front of house
Agents that talk to your shoppers. The good ones go past answering questions: they recommend products, build the order, track and cancel it, start a return, and hand over to you with the whole conversation when a human should step in. The difference from a chatbot is that the ticket ends resolved, not deflected.
02 · Back office
Agents that work on your data while you sleep. Each owns a domain (stock, margin, dead inventory, demand, revenue, the FAQ), scans on a schedule, writes up what it found and proposes what to do. You approve, it acts, and the result is measured against what actually happened.
03 · The shopper's side
Not yours, but here. AI shopping assistants now read product pages, feeds and policies on a buyer's behalf, and act on whichever copy of your facts they reach first. You don't deploy these; you make sure your store says the same thing everywhere they look.
Side by side
The line that matters runs from knowing to doing. Each kind of tool stops at a different point on it.
Reads your live store data
Decides without a rule written in advance
Changes things in your store
Finishes a customer's request end to end
Explains why it acted
Asks before touching money or a customer
Columns are categories, not named products, and a given product can sit between two of them. The last two rows are what a trustworthy agent should do: check them on any product you try.
Before you let one in
Hands are the point of an agent, and the risk. These questions separate one you can trust with your store from one you can't. Ask them of every vendor, including us.
How Lumnya does it
Lumnya is a team of ecommerce AI agents for Shopify stores: one at the front desk that sells and supports shoppers 24/7, and specialists in the back office that act on your data, with a memory that sharpens every week. Every action is confirm-gated and logged.
Back office
Seven specialists are live today, each with its own schedule and its own page. More are in training, and labelled that way until they ship.
Flags critical and low stock against your thresholds and opens reorder recommendations.
daily 06:00live
Computes true per-product margins from your costs.
daily 07:00live
Finds dead stock and proposes bounded clearance discounts you approve.
daily 08:00live
Mines your orders for products that really sell together.
Mondays 09:00live
Reads storefront conversations for demand your catalog didn't capture.
daily 07:00live
Decomposes every revenue move into quantified drivers.
Mondays 07:00live
Drafts the FAQ entries your assistant was missing, for one-click approval.
daily 08:00live
Front of house
The Customer AI recommends, builds orders, tracks and cancels them, runs returns, negotiates discounts under a ceiling you set and answers from your own knowledge base, in your shoppers' language, around the clock.
In control
Everything starts in recommend mode. Independence is set area by area, inside caps you choose; refunds, customer emails and return decisions never automate at all. Reversible actions undo in one click.
Where it runs
Shopify today, in private beta: free while the beta runs, no card. WooCommerce is on the roadmap.
Questions
An ecommerce agent is AI software that works inside an online store. It reads the store's data, decides what should happen next and acts on it, such as answering a shopper, flagging low stock or proposing a discount. A trustworthy one asks before anything touches money or a customer and logs every action.
No. A chatbot answers questions, and often its job is to keep the customer away from you. An ecommerce agent finishes the job: it finds the order, checks your policy, starts the return, and hands over to a human with the full context when it should. Agents also work in the back office, where no chatbot goes.
At the front of house: recommend products, build orders, track and cancel them, run returns and answer from your own policies, 24/7. In the back office: watch stock, compute true margins, find dead stock and propose clearance, spot products that sell together, read demand your catalog misses, explain revenue moves and fill gaps in your FAQ. Lumnya does all of these today on Shopify.
They can be, if they are built to ask first. Look for three things: anything touching money or a customer waits for your approval; every action is logged with its reason; and you can undo what is reversible. In Lumnya, every action is confirm-gated and logged, and refunds, customer emails and return decisions never run on autopilot.
It replaces the repetitive part of the work, not the judgment. The questions answered at 3 a.m., the nightly stock check and the Monday revenue read move to the agents; the decisions stay with you, because the agents propose and you approve.
Sidekick is Shopify's own assistant, built into the admin and free, and for help with tasks in the admin it may be all you need. Lumnya is a different shape: it talks to your shoppers on the storefront, runs specialist agents on a schedule in your back office, and keeps a memory of your store. The two can work side by side.
Nothing while the private beta runs: the whole product, free, with no card, no plan and no monthly conversation limit. Launch pricing is announced before general availability, and merchants who join during the beta keep a founding rate.
Lumnya runs on Shopify today. WooCommerce is on the roadmap; it gets a date on the roadmap page when it has one, not before.
Private beta
Join the private beta and be first in line when Lumnya opens to your store.