AI for e-commerce brands, from the order to the review

An online shop runs on a dozen tools that never talk: Shopify or WooCommerce for the orders, Stripe for the money, Gorgias or Zendesk for the tickets, Klaviyo for the email, Meta and Google for the ads, Trustpilot for the verdict. Memtro connects to all of them and answers the questions that cross them, with every figure cited to the order, charge or ticket it came from.

Start for free Read the docs

7-day free trial · card on file, nothing charged until it ends · cancel any time

Orders, customers, stock, support tickets, ad spend and the ledger answered from one memory, with every figure cited.

The systems these questions draw on, connected to one memory. Every connector reads by default; the tools that change something are granted by name, per team.

The problem

What it looks like today

01A support ticket about an order means opening the shop, the payment processor and the carrier before replying.
02Campaign spend sits in two ad managers and never meets the orders it produced.
03The same customer appears in the shop, the help desk and the email tool under three records.
04Reviews and social comments are noticed days late, after the next customer has read them.
05Stock, returns and the carrier each tell a different story about the same order, and support picks one.
06Nobody notices that a product's returns have doubled until the month's figures are in.

How it works

What Memtro does about it

One memory in front of the systems you already pay for, reading by default and acting only where an administrator has granted the tool by name.

Connect Shopify or WooCommerce, Stripe, the help desk, Klaviyo, the ad platforms, Google Analytics and Trustpilot, reading by default and acting where you grant it.
Ask about a ticket and the agent reads it, finds the customer's orders and charges, and answers with each fact cited; Gmail can draft the reply when asked.
Ask how a campaign did and the answer joins spend from the ad platform to sessions and orders, with cost per order worked out and shown.
Customers, orders and tickets are anchored into one memory, so the second question about a customer is faster than the first.
Schedule jobs: failed payments every hour, new reviews under four stars each morning, campaigns whose cost per order doubled this week.

The stack

System by system

What Memtro reads in each part of the stack, and what it may change there once the tool has been granted.

Carriers and delivery

ShipEngine or EasyPost in front of Royal Mail, Parcelforce, DPD, Evri, InPost and Yodel, or Royal Mail Click & Drop and UPS directly on your own accounts: what a parcel costs before the postage is bought, where it is afterwards, which ones have stopped moving, and the lockers and ParcelShops nearest a customer. Buying a label, voiding one and booking a collection are each granted by name, and a price is always quoted before anything is spent.

The shop

Shopify or WooCommerce: orders, line items, fulfilment status, customers, products and inventory, with the order a customer is asking about found by number, email or name.

Money

Stripe, PayPal or Square: charges, refunds, disputes and payouts, matched to the order that produced them and the customer who placed it.

The help desk

Zendesk, Intercom, Front, Help Scout, Freshdesk or Gorgias: the conversation, that customer's history, and — on the desks that offer those tools — a reply or a status change when you ask and it has been granted.

Email and flows

Klaviyo, Mailchimp, Brevo or Customer.io: campaigns, flows, sends, opens and clicks, so a spike in orders can be traced back to the send that caused it.

Advertising and analytics

Meta Ads, Google Ads, TikTok Ads, Microsoft Advertising, Google Analytics 4 and Search Console: spend and sessions on one side, orders on the other, joined in the answer rather than in a spreadsheet.

Reputation

Trustpilot, Google Business Profile, Meta Pages and Instagram: what customers said in public this week, read every morning instead of when the next one mentions it.

There are 144 connectors in all, each with its own tools, its own permissions and a page of its own. Browse every connector →

Who it is for

Who asks what

Founder

Asks the week's question once: what sold, what it cost to sell, what people said about it, and which of those three has moved since last week.

Support lead

Answers a ticket with the order, the charge and the delivery status already in the reply, and sees which customers have more than one thing going wrong at once.

Performance marketer

Joins spend to orders without exporting anything: cost per order by campaign, this week against the last four, each figure cited to the platform it came from.

Operations

Watches what breaks: failed payments, orders paid and unshipped, and products whose returns or complaints are rising.

The work that comes round

A week in the shop

01
Every morning
The overnight jobs have already run: failed and disputed payments in the last hour, new reviews under four stars, and the tickets attached to orders that have not shipped.
02
Through the day
A ticket is answered with the order, the charge and the carrier status in the same reply, drafted where you have granted the help desk or Gmail action and sent by a person.
03
Monday
Spend, orders and cost per order by channel for last week against the week before, posted to the channel the team already reads rather than assembled by hand.
04
Over time
Customers, orders and tickets are anchored into one memory, so the second question about a customer starts where the first finished, and what was decided about a supplier or a discount is still there in March.

In practice

Ask it like this

Plain English, from chat, your editor, a scheduled job or your own agent. In agent mode each question also checks the other systems that know about the same people, companies and references, and the answer ends with a line per system consulted.

How agent mode works Watch the tour

Memtro · agent mode

Why is order #48213 late, and has the customer already written in about it?

Reading 5 systems…

Shopify
WooCommerce
Stripe
ShipEngine
Royal Mail Click & Drop

Every figure in the answer cites the record it came from, and the run ends with a line per system consulted.

›How did last week's Meta and Google campaigns do, and what did each order cost?
›Which customers have more than one open ticket and an order waiting to ship?
›What are reviewers saying about delivery this month?
›Which products had the most returns or complaints this month, and what reason did customers give?
›Did Thursday's Klaviyo campaign move orders, and for which products?
›Which customers have spent over £1,000 this year and have an open ticket?
›What is the cost per order on Meta this week against the last four weeks?
›Where is order 10492, and has the carrier scanned it since it left us?
›Which parcels sent over five days ago have had no carrier scan since?
›What is the cheapest service for a 1.2kg parcel to Bristol today, and what would next-day cost?
›What did we spend on postage last month, by carrier?

On a schedule

Work that runs itself

A job is a prompt the agent runs on a cron schedule as its owner, with their tools and connections. Say any of these in chat and the job exists; every run's report is kept in a chat you can question.

schedule:Every hour: failed or disputed Stripe payments in the last hour, with the order and customer
schedule:Every day 09:00: new reviews under four stars and social comments that need a reply
schedule:Every Monday 08:00: spend, orders and cost per order by channel, last week against the week before
schedule:Every day 07:00: orders paid more than 48 hours ago and still unfulfilled, with the customer's last message
schedule:Every Friday 16:00: products whose refund or return rate rose this week, with the reasons customers gave
schedule:Every day 09:00: parcels with no carrier scan in 48 hours, by carrier, with the customer and the order
schedule:Every Monday 08:00: postage spend by carrier and service against the week before

How jobs work

Questions

Questions people ask

Can it refund or change an order?+

Only where you grant it. Help desk actions such as replying on a ticket or changing its status are granted by name per team and run only when a person asks; refunds and order changes are granted the same way as those actions arrive. Nothing changes in the shop until you say so.

Does it work with both Shopify and WooCommerce?+

Yes, and with either on its own. Orders and customers from each are anchored into the same memory, so a brand that runs two storefronts asks one question.

How current is the data?+

Connectors sync hourly and the agent reads live when you ask, so a question about an order placed ten minutes ago is answered from the shop, not from a stale copy.

Can it see stock?+

It reads what the shop records: inventory levels and product data in Shopify or WooCommerce. A warehouse or 3PL system that exposes a database is reachable over Postgres, MySQL or SQL Server, read-only, with the query shown in the answer.

Does it handle several brands or storefronts?+

Yes. Each shop, payment account and help desk is its own connection, shared with the team that runs that brand, and a group question is answered across all of them with the store named on every figure.

What does it cost to run?+

One subscription per organisation by headcount, with connectors, jobs and memory never metered. Model usage is billed by the providers whose keys you bring, and routing sends an order lookup to a cheap model and an investigation to a strong one, so the bill follows the question.

The same work, by team

The same connectors and the same memory, described through the job somebody does rather than the business they do it in.

Memtro for e-commerce brands

Ask it about your own business

Connect the first two systems, ask the question you opened this page with, and see where the answer comes from. About five minutes.

Start for free Pricing