Cross-border work has a shape: **high volume, highly repetitive, and expensive when it goes wrong**. Creator outreach means paging through the back office until your hand hurts, and a repeat invite costs you account standing. Support gets a message in Thai at 2am. Product research is done by hand. These are the lines we have actually run.
Free credits on signup · No credit card · Mac and Windows
Page through the back office, tick boxes, send, write it down. The thing you dread is the repeat invite — the same creator getting asked again and again costs you account standing. And the list you keep locally rarely matches what the platform actually shows.
One message in Thai, one in Vietnamese, one in English. Translation gets you the gist, but who decides when a refund is approved, or when a replacement ships? Every agent decides by feel, and the more languages you add the further the answers drift apart.
Scroll the charts, check competitors, mine the reviews, put it in a sheet. Then remake every asset at the right size and length for each platform, once per language. Finish one product and the next one starts from zero.
One set of numbers in the platform back office, one in your own ledger, one in the ads console. You find the gap when you stitch them together at month end, by which point tracing it to an order is too late.
Each of these has run inside a real seller's process. Open one to see how it moves and where it stops for a person.
Candidates filtered by category and performance, then deduplicated three ways — your local ledger, platform state, and invite history. You get a preview. What you approve is exactly what goes out; it never tops the list back up to fill a quota. Afterwards it reconciles against what the platform shows.
The rules for refunds, replacements and delivery exceptions live in one place and every language shares them. Anything involving money is never executed automatically — it assembles the reasoning and the evidence and waits for you.
Multilingual customer service: from slow, inconsistent replies to one line with a standard →
Through the API where one exists, and through a real browser carrying your signed-in session where it does not. Both routes pass the same approval gate.
Research turns charts, competitors and reviews into a sheet you can actually compare. Creative produces images, video, voiceover and subtitles per SKU and platform, with the product itself locked to a baseline shot — the model only handles scene and camera.
Platform, shop and ads numbers reconciled side by side, with gaps traced to orders. When you run several shops across several platforms, one command center ties the lines together.
Where there is no public API it uses a real browser on your own machine with your own signed-in session, at human pace rather than scripted clicking. Where an official API exists, that comes first.
Fetching and execution happen on the machine with the desktop app installed. Only the model call you chose leaves the network, and what leaves is the task description, not your shop data. You can route that call through your own model as well.
Yes. Each shop is its own session with its own credentials, and definitions can differ per shop.
Free credits on signup, then usage-based, with individual and company plans. Cross-border sellers usually take the USD subscription.
Plain language is enough — "show me creators in this category we have not invited in the last 30 days". What actually needs configuring is your own definitions (what counts as a qualified creator, when a refund can be approved), and that gets calibrated at delivery.
Free credits on signup, macOS and Windows. If you would rather see it run first, book a walkthrough and we will take it through your real process.
Free credits on signup · No credit card · Mac and Windows