Our promises · v1 · 16 Oct 2026
The feed that ends.
Converge borrows the pull of a short-video feed and refuses its traps. These eight commitments are public so you can hold us to them. If we break one, tell us through Feedback, in the menu on every page.
- 01
Finite decks with a real end
Every session is a short deck of cards with a counter and an end screen: what you learned, what you built, what is due next. The next deck is shown as a choice. It never auto-plays.
The EU Commission's 2026 finding against TikTok named infinite scroll and autoplay. We keep the pull and drop the trap.
- 02
We show our working
Lessons are picked for what you need next: what you're weakest at, what's due for review, your goal, and what comes next on your path. Never watch time. "Why these lessons?" shows the numbers for every tile, and one tap switches to plain path order.
You should be able to check what a feed is built to do.
- 03
Stopping cues
After 2 decks in a row, or after 22:30 local time, you get a full-screen "good place to stop" card that sums up your progress. It is a real exit, not a nag.
Context-aware break prompts work; spaced practice beats cramming.
- 04
No guilt, no paid repair
Streaks are weekly (3 decks or one shipped task), with 2 free freezes and rest built in. There is no paid streak restore. Notifications are only ones you schedule, at most 1 a day, quiet 22:00–07:00 by default.
Breaking a strict streak causes drop-off. Guilt burns people out.
- 05
Private by default
Your profile and your leaderboard row are hidden until you switch them on. Projects are made public one at a time. Learners aged 16 or 17 keep both off unless they confirm, their crew posts are for members only, and they get no late-night reminders. You can download everything as a JSON file or delete your account from Settings.
Showing your work should be a choice you make per item.
- 06
AI help is labelled
Lessons are written with AI help, checked against sources, and say the date they were last fact-checked. Facts about models and prices cite official docs.
People's top ask of platforms is to label AI content. It is also just honest.
- 07
We check it's good for you
A one-tap "was this time well spent?" sits in the feedback widget on every page. We track regret and late-night use next to retention, and change things when the answers are bad.
What you measure is what you end up optimising.
- 08
Outcomes, not minutes
We measure what you remember (how many flashcards you still know) and the projects that pass our checks, not daily active users. Watching earns 1 XP; a first-try fix earns 18.
Points for showing up undermine interest. Points for quality do not.
What we never do
- Loot boxes or random rewards for points. Surprise comes from the lessons and the story, never from prizes.
- Paid boosts, paid streak repair or paid rank. A plan buys full access to lessons and tools; nothing buys rank.
- Selling your data or showing ads.
- Ranking the public leaderboard on anything but checked work and a small effort term.