When an AI conversation recommends a tool or product, Retriever tells you if cashback is available. Merchants pay us a commission when you sign up through an offer you clicked; we split it with you. Your conversations and your code never leave your machine.
Free. No card, no account to start. The commission split is the whole business model, stated upfront.
Every recommendation you follow earns someone a commission. Retriever routes that money to you, with three things nobody else has.
We take the affiliate commission (percentage, fixed bounty, or recurring SaaS revenue share) and split it with you. Recurring commissions become recurring cashback: money back every month you stay subscribed.
The offer catalog syncs to your machine, and your machine decides what is relevant: the pages you read and the projects you work on are never transmitted. That claim is checkable, because the client is open source.
In your AI conversations, in your editor while your agent works, and soon wherever recommendations happen. Offers appear where decisions are made, never as noise.
Nothing activates without your click, and every state of your money is visible.
Add the extension, or run npx retrievercash init for the developer setup. Backup kept, one-command uninstall.
A discreet chip shows the cashback amount when a partner is mentioned. Interested? Click, and only then does an affiliate link exist. Not interested? Dismiss it. No auto-cookies, no background anything.
A purchase shows up as tracked with its estimated amount, locks as pending with a real countdown (the merchant's validation window), then becomes available once the network confirms and pays. Refunded on the merchant side? The entry is cancelled, with the reason shown.
The cashback extension category earned its bad reputation. We exist because of that, not despite it. Security →
We claim a commission only when you actually clicked an offer we showed you. No affiliate cookies dropped in the background, no last-click hijacking of creators' links, no hidden better coupons.
Your conversations, your browsing, your code, your dependency list. Detection runs locally against the synced catalog. Our server sees an anonymous device id, the offers you clicked, and the conversions merchants report. That is the whole list.
Every line that runs on your machine is public. All network calls live in code you can read in one sitting.
Pending amounts are estimates until the merchant's network confirms; we pay out after we are paid, via regulated rails. No float games, no instant-cashback marketing.
Your offer reaches people at the exact moment an AI recommends your category, targeted by real context without collecting a byte of their data.
| Classic cashback portals | Retriever | |
|---|---|---|
| Audience | Deal hunters at checkout | People at the moment of recommendation |
| Targeting | None | Context, matched on-device |
| Attribution | Last-click disputes | Commission claimed on shown-and-clicked offers only |
| Pricing | CPA | CPA, including recurring revenue share |
Programs are onboarded manually while the network is small. Write to partners@retrievercash.com.
It depends on what you adopt. Amounts are shown per offer before you click, and SaaS revenue-share programs turn into recurring monthly cashback. We publish no fantasy numbers.
Yes, and we say so. Cashback has always been affiliate commission shared with the buyer. The difference is who gets profiled (nobody: detection is local), how activation happens (only your click), and how much of the pipeline you can audit (all of the client).
Merchants have a validation window (typically 30 to 90 days) to account for refunds. Your entry shows the exact unlock date. We release money once the network confirms and pays.
No. Detection runs entirely on your device against the cached offer catalog. Nothing you read, browse or write is transmitted, and the client source proves it.