This is a repost. Originally published on the Arvio blog: https://arvio.a.xyz/blog/what-arvio-actually-does. The canonical URL points back there.
Key takeaways
- You do not have to know what to ask for. Arvio proposes a goal for the week from a conversation with you — or you set your own — then tells you, each day, which job moves you towards it. It proposes; you decide. Most tools start one step later, with a list you have to prioritise yourself.
- Most tools hand you a to-do list. Arvio drafts the work behind it — and does it the moment you approve. Describe a change in plain language, review exactly what it will do, approve once.
- One sentence, one approval, the whole job. In one recorded run, a single sentence repriced a live 69-product, 153-variant catalogue in 5 minutes 46 seconds — and spent 2 minutes 17 seconds of that showing the plan before writing anything.
- You stay in control. Nothing reaches your store until you approve it. Arvio never deletes anything, and never touches your theme.
- It tells you what to fix first. It reads your store, works out which stage it is at, and ranks the work by what is blocking the next one. A single-purpose tool is better than us at its one thing; what it cannot do is tell you whether that thing is this week's thing.
- Built by Adot, whose previous product AiLead has been installed by more than 3,000 e-commerce brands to date. Arvio itself launched on the Shopify App Store in April 2026.
Written for partners as much as for merchants. If you build a Shopify app, run a roundup or review site, or promote apps for commission, the terms and the offer are here — including what we can and cannot answer yet.
This page describes it. The listing is where you install it
Arvio is an AI store operator for Shopify: it reads your live catalogue, proposes what to change, and applies nothing you have not approved.
What Arvio is
An AI store operator for Shopify. You say what you want in plain language — rewrite the descriptions for everything in this collection, raise every price by 10%, fix the missing alt text — and Arvio prepares the work, shows you exactly what it is about to change, and executes after you approve.
The difference from most AI tools is where the work stops. An audit tells you 400 product pages are missing meta descriptions. A dashboard tells you conversion slipped. Someone still has to open 400 products and type. Arvio is the part after the report.
It also starts one step earlier than a report does. Before you ask for anything, Arvio will suggest what this week should be about — from what it reads in your store and from talking to you — and you either take that goal or set your own. Everything it proposes afterwards is measured against the goal you agreed on, which is why the same store gets a different list in different weeks.
Who is behind it
Arvio is built by Adot, a team working on AI and cross-border e-commerce. Our other product, AiLead — a guided-selling AI agent that works on the storefront side — has been installed by more than 3,000 e-commerce brands to date — that is the cumulative install count from our own dashboard, stores currently running it and stores that have since uninstalled — and Arvio is the same team applying that to the back office.
Our founder, Dr. Wei, holds a PhD in AI and was one of the earliest members of Google's mobile ads team. Arvio itself launched on the Shopify App Store in April 2026.
Who it is for
Stores where the owner is also the operations team — a few dozen to a few thousand products, one person doing copy, SEO, pricing, email and support.
That constraint decides the design. There is nobody to delegate to and no time to learn a tool, so every job is one instruction and one approval, and the work happens whether or not you know how the bulk editor works.
What it does
Product content. Rewrite product descriptions and titles across a collection or the whole catalogue. Fill in SEO titles, meta descriptions and image alt text. Draft copy for products that have none.
Pricing and inventory. Change prices by percentage or by rule, with a per-variant margin check. Adjust inventory selectively. Create discount codes.
Store SEO. A store check that ranks issues by what they are costing you, plus a technical health score covering page speed, whether robots.txt is blocking your site, and whether your sitemap is valid.
Analysis. Conversion-path diagnosis, traffic-source analysis, and competitor pages fetched and compared — so "look at these three competitors and tell me where my prices sit" is one job rather than three tools.
Marketing email. Draft campaigns in your own Klaviyo account, for you to review and send — same approval step as everything else.
Two things it deliberately will not do: it does not delete — asked to remove a product, it sets the product to draft — and it does not touch theme files or write code.
What it is not
Not a customer-facing chatbot — it never talks to your shoppers. Not an autonomous sender: it drafts, you send. Not a fulfilment, 3PL, POS or payment-disputes tool. Not a translation product. Not a theme or page builder. Not a reporting product — if what you want is a dashboard, buy a dashboard.
Watch it work
A four-minute walkthrough of the whole thing on a live store — the diagnosis, one plain-language instruction, the approval card, and one of the tasks that can be undone afterwards:
Arvio: AI store operator — full overview (4:16)
If you would rather read than watch, the rest of this page covers the same ground — the run below is the same kind of job, written out step by step.
One run, start to finish
We recorded Arvio doing a real job on our own live Shopify store — 69 products, 153 variants — so you can watch it rather than take our description of it. Sixteen seconds of that run is public: prices going in by hand, then the same job as one sentence, with the confirmation card in between. Everything you see is our own store, and the blurred boxes are ours too — account details, plan and connection status. It is our store and it is one store; the full write-ups linked below say what that does and does not establish.
The instruction was one sentence: increase the price of every product by 10%. End to end it took 5 minutes 46 seconds, and 2 minutes 17 seconds of that was spent preparing a preview before anything was written.
That preview is the product. It is not a yes/no dialog — it lays out the scope, the currency, which variants it could check margin on, and the named old → new prices:
Nothing above is written yet. The button is the moment it becomes real.
It also caught something we had not asked about — in the preview, before anything was written. Three sizes of one basket were already priced out of order: the bigger basket was cheaper than the smaller one before we touched it. A flat 10% preserves an inversion rather than fixing it, so Arvio proposed moving those three by more than the 10% we asked for, said so on the card, and waited.
That is the part worth being precise about, because "the agent exceeded my instruction" is a reason to distrust a tool, not to buy one. It did not exceed anything: it put the exception in front of us, named it, and we approved it. Had we wanted the literal 10%, the card was where we would have said so.
A second run on the same store: 23 products had no description at all. Worth noting what these two runs say next to each other — repricing 153 variants took 5 minutes 46 seconds and drafting 23 descriptions took 5 minutes 23 seconds, so on our evidence the time is mostly fixed cost rather than per-product. Two runs is two runs; we are not turning that into a throughput curve. Asked to draft descriptions and show them first, it produced all 23 in 5 minutes 23 seconds and saved none of them — the drafts waited for approval, exactly as asked.
Both runs are written up step by step, including the parts that surprised us: the price run and the description run.
What day one looks like
Install takes a click and no code. Arvio then reads your store and comes back with a score and the issues holding it back, ranked. You pick the top one — or ignore the list and ask for something else in your own words. The two jobs we would start with are a batch of product descriptions and the missing SEO fields — those are the ones where you can open a product afterwards and see the change.
After that first job, Arvio settles into the weekly rhythm described below: a goal you agreed to, and one recommended job waiting for you each time you open it.
How it decides what to fix first
An early store's real problem is rarely "I cannot do this task". It is "I do not know which task matters this week".
Arvio reads your store against four stages — Store Ready → First Sale → Stable Sales → Growing — works out which one you are in, and ranks what is blocking the next one. The tasks it proposes come out of that, which is why the same request produces different priorities on two different stores.
That judgement is not a one-off scan. It becomes a goal for the week. Arvio suggests one from what it has read and from what you tell it you care about; you can accept it, change it, or set a goal of your own. Then it looks at the store each day and tells you which job to do next against that goal — not a fresh list of everything wrong with your catalogue, but the one thing that moves this week's number. It waits for you inside Arvio; it does not chase you. Skip three days and you have not fallen behind a queue — the next time you open it, the recommendation is the one that fits the store as it is that morning.
Two things worth being exact about, because this is the part most easily oversold. Arvio proposes; it does not act on its own. The daily output is a recommendation with the work already drafted behind it — you still approve before anything reaches your store. And the goal is yours to overrule: if you decide this week is about something else, say so and the ranking changes with it.
It is also the honest answer to "how is this different from an app that does one thing well?" A single-purpose tool is better than us at its one thing. What it cannot do is tell you whether that thing is what you should be spending Tuesday on — and it cannot do the next thing, or the one after, without you buying and learning two more tools. The trade is depth for coverage plus a decision about order. If depth on one job is what you need, buy the specialist; we will tell you the same thing if you ask us.
Everything above, on your own store
The fastest way to see whether this fits your catalogue is to point it at your catalogue. Every change waits for your approval, and anything it did can be undone.
Control
- Nothing goes live without you. Every change is drafted, itemised and shown before it runs.
- One click approves that one job, and it ships across your store — with the reasoning attached to each action. Approval is per job: it does not carry over to the next one.
- A run can be rolled back on the tasks that support it, restoring the values that run changed. The supported list is still growing, so it is in the request list below rather than on this page — we would rather hand you a current one than print one that goes stale.
- Access is OAuth, revoked from inside Shopify — which is table stakes, not a feature; we list it only so the question is answered.
What it connects to
The right-hand column is what we are building, without dates on it. If you are writing about us, quote the left one.
| Working today | Coming next |
|---|---|
| Shopify Admin (read and write) · Shopify Checkout · Klaviyo · Google Search results, read for competitor and ranking checks | Google Ads · Google Analytics · Meta Ads · TikTok Ads · Judge.me · Yotpo · Mailchimp · Gorgias, and more after those |
Languages
The admin, and everything Arvio writes, in English, German, Spanish, French and Italian.
Pricing
Three plans, each with a 5-day free trial: Starter $9.90, Growth $39, Pro $99 per 30 days. Work is metered in credits — a completed run costs 10 credits, and the plans carry 800, 4,000 and 10,000, resetting every 30 days — about 80, 400 and 1,000 runs. A run is one instruction, not one product: repricing those 69 products was a single run, and so was drafting 23 descriptions. Failed or incomplete runs are not charged. To put the credits in terms of use: if you act on one suggestion a day, Starter's 80 runs cover a month with room to spare.
There is a practical ceiling on how much a single run can do. We would rather tell you where our evidence stops than print a number we have not tested: the largest run we have recorded is 69 products and 153 variants, done as one run. We have not published a run at several thousand products, and we do not know where the ceiling sits between those two points. If your catalogue is in that gap, ask us before you buy — hi@a.xyz — and you can split any job by collection to keep each run smaller.
FAQ
Does it work on a large catalogue?
One product, one collection or the whole catalogue in a single run, and you can ask it to go collection by collection if you would rather approve in smaller batches. Our own recorded run was 69 products and 153 variants; we have not published a run at several thousand.
Does Arvio change anything on its own?
No. Every change is drafted and itemised first, and nothing is written until you click Confirm on that job. Approval is per job — it does not carry over to the next one. Arvio also never deletes anything and never touches your theme: asked to delete a product, it unpublishes it to draft instead.
Can I undo a run?
On the tasks that support rollback, yes — it restores the values that run changed. The supported list is still growing, so we keep a current one to hand rather than printing one here that goes stale; ask for it at hi@a.xyz.
What does a run cost?
A completed run costs 10 credits, and a run is one instruction, not one product: repricing 69 products was a single run, and so was drafting 23 descriptions. Starter carries 800 credits per 30 days — about 80 runs. Failed or incomplete runs are not charged.
Where we are taking it
The stage model is also the roadmap. Three things we are building next:
- Turning visits into orders — add-to-cart and checkout diagnosis, with product page fixes ranked by what they are costing.
- Bringing customers back — win-back and repeat-purchase campaigns built from what your store data already shows.
- Knowing what worked — the weekly goal and the daily ranking are running today; what is not there yet is the loop closing on itself: measuring the outcome of the actions Arvio ran and feeding that back, so next week's ranking is better than this week's because of what happened, not only because of what the store looks like now.
Working with us
Copy-paste summary for editors. If you are writing us up, this is the description we stand behind — take it verbatim and we will not ask you to change it later:
Arvio is an AI store operator for Shopify, built by Adot Technologies Inc. It reads a live catalogue, proposes what to change, and applies nothing the merchant has not approved. Describe a change in plain language, review exactly what it will do, approve once — then it runs across the whole catalogue. It never deletes anything and never touches the theme. Plans start at $9.90 per 30 days after a 5-day free trial. apps.shopify.com/arvio
If you build a Shopify app. We execute broadly across the admin; most apps go deep on one job. In practice that means two things worth talking about: when a merchant asks us for something your app does properly, we would rather name you than half-do it — and when your app produces a list of findings that someone then has to action one by one, that hand-off is exactly what we are for. Tell us which category you own and we will stay out of it, in writing if you want it in writing.
If you run a marketplace, roundup or review site. Categorise us as an AI store operator — the thing that does the catalogue work — rather than as ops automation for teams. The run write-ups linked below carry the methods and the parts that did not reproduce, and the left-hand column of the integrations table is the one to quote — the right-hand column is unshipped and undated.
If you promote apps for commission. The headline terms: 50% of net recurring revenue for 12 months from each merchant you refer, recurring subscription revenue only, run through Shoffi. Entry is $9.90/30 days after a 5-day trial, so it is a low ticket with a long tail rather than a big first payout.
That is not enough to price it, and we would rather say so than let you find out after the email. Three inputs you need and this page does not have: what "net" excludes, the attribution window and payout threshold (both set on the Shoffi side), and trial-to-paid conversion. The first two are in the request list below. The third we are not publishing — the base is too small for the number to mean anything, and you should discount any four-month-old app that hands you one.
What we will send you
One address for all of it — hi@a.xyz — and this is the whole list, so you know what is worth asking for:
- the full recordings of both runs, not the sixteen-second cut
- the current list of tasks that support rollback
- what "net" excludes, plus the attribution window and payout threshold, as Shoffi defines them
- screenshots, or a demo store to try it on
- a written category exemption, and a straight answer on whether it is enforced in the product or by us
Related: How to change prices in bulk in Shopify · How to bulk edit product descriptions in Shopify
Published 26 August 2026 by Adot Technologies Inc, the team behind Arvio: AI Store Operator on the Shopify App Store. Questions, or something here that needs correcting: hi@a.xyz.
Arvio: AI Store Operator — install it on the Shopify App Store. Everything on this page, running against your own catalogue.
Originally published at https://arvio.a.xyz/blog/what-arvio-actually-does. More Shopify bulk-editing writeups are on the Arvio blog.
Top comments (0)