About Feedpik
We think you should own your news algorithm.
Feedpik is an AI RSS feed reader built on a single conviction: the thing deciding what you read all day should be filtering for you — and yours to argue with and change.
Three shapes, none of them yours
Reading on the open web settled into three shapes. The engagement feed is the one most people use, and it works — it is ranked well, and it will often put something in front of you that you are glad to have seen. It is just not ranked for you. It is ranked for someone else’s revenue, and on the days those two part ways, you do not get a vote.
The chronological timeline is the honest answer, and it stops being usable past a few dozen sources, because recency is not the same as importance. The products that did offer control mostly handed you a rules engine and wished you luck — trading a black box for a second job. We did not think those were the only three shapes available.
So we built a better option
Feedpik is a decentralized personal information system. RSS had already decentralized half of it — no company decides which sources may reach you. Feedpik does the same to the other half, the ranking. It runs on our servers, but it is not ours: it is yours to change whenever it gets you wrong, and nobody here tunes it for you. There is no house algorithm that everybody shares.
AI is what keeps that from being a chore. It goes and finds the sources worth adding, and it retunes the ranking when you tell it, in plain words, what you want more and less of — so having your own algorithm costs you a sentence rather than an afternoon.
There is further to go. Running the ranking on your own machine, local models, local storage and end-to-end encrypted preferences are the direction, because the fewer of your decisions that have to sit on our servers, the less any of this rests on us keeping our word.
What Feedpik is built on
The algorithm should answer to you
The job of an algorithm is to find the handful of things you actually want out of the thousand you do not. Feedpik does that job for you — and when it gets your priorities wrong, you change them, in plain language or with a dial, and every change is a version you can roll back to. It answers to you, not to an ad market.
Control should not mean homework
Most tunable products quietly make you the algorithm engineer. That is a worse deal than the black box, not a better one. So Feedpik takes instructions in plain language, and it takes them from your AI agent too — hand over a scoped key and let something that already knows your taste drive. Agents propose; you own the result.
Your reading is for your ranking, nothing else
The usual deal is that a feed studies what you read so it can describe you to advertisers — the sharper the description, the more you are worth. Feedpik studies what you read for the opposite reason: it is the only way to rank for you rather than for an average. No ad profile, because we sell no advertising. No analytics of any kind. Nothing sold or rented, ever. Same data, one job.
You are not one audience
Every other feed builds a single profile of you and serves it back — the same you at work, on a Sunday, and three years from now. The person researching a client has little in common with the one looking for something to read in bed. Feedpik gives each of them a persona: its own sources, its own ranking, its own record of what you liked there. Switching swaps the whole feed, not a filter on one.
Your own information system
The sources you picked, an algorithm that is yours alone, your AI agent filtering again on top, delivered the way you like to read.