













Ever tapped "Join" on a favorite creator's channel? Those $4.99 memberships are subscriptions too and the easiest ones to forget you're paying.
What's new in v1.0.4, rolling out this week:
App Store & Google Play subscriptions detected automatically
CentryAI now scans your iCloud and Gmail for App Store and Google Play receipts surfacing subscriptions you may have forgotten, complete with app icons, trial prices, renewal prices, and next billing dates.

i signed up mostly out of curiosity after seeing the tagline (forgetting 11 subscriptions is extremely relatable) and it actually found a couple i'd genuinely lost track of, an old vpn and a design tool trial that quietly converted to paid. the renewal alerts are timed early enough that you can actually cancel before the charge instead of after, which is the whole point and something a lot of similar trackers get wrong by pinging you the same day.
the auto-detection from email/inbox scanning is good but it missed a subscription that bills through apple's app store since there's no email receipt to scan, i had to add that one manually. would be nice if it could also pull from the app store/play store subscription list directly instead of relying purely on inbox parsing.
looked at just tracking things manually in a spreadsheet again, which is what got me into this mess in the first place since i never actually update it. also considered the subscription tracking feature bundled into my bank's app, but that one only sees charges after they've already hit, so there's no lead time to cancel anything.

@yilmazemreΒ Congrats on the launch! This feels super relatable. Also, wanted to know how Cancel Finder actually locates the exact cancellation page, is that a maintained list per subscription, or does it search/navigate live each time?
@vijay_garg5Β Thanks! It's a hybrid of both. There's a curated catalog of direct cancellation URLs for the most common services β those resolve instantly, no searching. For services not in the catalog (or when a page has moved), it resolves live, and every URL is validated before the app opens it β no open redirects.
The part I like most is how it stays fresh: after you use a link, the app asks "did this work?" Downvoted URLs get invalidated and re-resolved; upvoted ones get promoted. So when a service quietly moves its cancel page, the first person who hits the dead end heals it for everyone after them.
The inbox-scan then detect-recurring-charges then score-what-you-don't-use flow is the useful core, but connecting Gmail/iCloud is also the part I'd want to understand before granting access: does the email parsing run on-device, or does my mail get sent to a server to find the charges? And is it a one-time scan I can revoke, or a live connection that keeps reading new mail? For a utility like this, where the parsing actually runs is basically the whole trust decision.
@hi_i_am_mimoΒ Great question β where the parsing runs is exactly the right thing to ask before granting access. Full transparency:
The parsing runs server-side, not on-device β but your emails are processed in memory and never stored. The only thing that gets saved is what you confirm: service name, price, renewal date. Not the emails themselves, not even snippets.
It's a one-time scan per tap, not a live connection. Access is read-only, the scan runs only when you press the button, and we don't hold tokens that would let us read your mail in the background. You can revoke access anytime from your Google Account β Security β Third-party apps.
On the trust side: CentryAI is a Google-verified app β to keep Gmail access we went through CASA Tier 2, an independent security assessment Google requires for apps touching Gmail (penetration testing, code scanning, the works). We passed it this year and it's re-validated annually. iCloud works differently: an app-specific password that lives only in your device's Keychain and is wiped on sign-out.
Happy to go deeper on any part of this β it's the piece of the app I'm most opinionated about.
Nice β the "find the cancel page in one tap" piece is the part that actually matters.
A lot of those forgotten subscriptions aren't just forgotten, they're engineered to be: negative-option billing, a cancel flow buried three menus deep, the "are you sure?" guilt screen.
The FTC's whole click-to-cancel push exists because companies bank on that friction. Anything that collapses cancellation back to one tap is doing real consumer-protection work. Curious whether you flag the free trials that auto-convert before the first charge lands β that's where most people actually get caught.
@peterdigitalisΒ Exactly right β the friction is the business model for a lot of these companies. Negative-option billing, the three-menu-deep cancel flow, the guilt screen asking if you're really sure β none of that is accidental UX. It's revenue retention by design.
That's precisely what Cancel Finder is built to cut through. One tap, straight to the cancel page, no detours.
@yilmazemreΒ Revenue retention by design β that's the perfect phrase for it, might steal it. And if Cancel Finder ever grows a trial-watch mode β flag the auto-convert date the moment the confirmation email lands, before the first charge β that's the version I'd be recommending to people. It's the one moment where a reminder beats a refund fight. Good luck with the launch π
@peterdigitalisΒ Steal away π And noted β trial-watch mode is now officially the most-requested feature of launch week. When it ships, you'll have named it.
Cancel Finder gets the spotlight but detection recall is the harder problem. Annual and biannual charges only surface once in the whole mailbox, so a one-time scan can miss the exact subscriptions that hurt most, the $120/yr thing you forgot a year ago. When I built inbox parsing the real killers were bundled receipts (Apple emailing five app subs in one 'your receipt'), forwarded receipts, and non-English merchants. How do you handle the annual ones and the Apple/Google bundles?
@dipankar_sarkarΒ You clearly built one of these β that's exactly the hard-problem list π Straight answers:
Annuals: you're right, and I won't pretend otherwise. The current lookback is ~90 days β a deliberate first-version tradeoff (fast scans, minimal inbox reading). So yes, the $120/yr charge from ten months ago is missed today. Extending the lookback is high on the roadmap β you're the third person to raise it this week, and suggestions like yours are exactly what's shaping how we build it.
Bundles: Apple/Google/Amazon are treated as platform billers β the sender domain is never trusted as the service. The extraction step is designed to produce one entry per app from a single bundled receipt, each with its own name and amount.
Forwarded + non-English: both handled β extraction explicitly looks for "receipt from X" in forwarded mail, and detection works across languages (the app ships in 18).
We deliberately biased toward precision first (false positives kill trust faster than misses) and we keep rebuilding recall service-by-service via community-learned patterns. Very much still improving β and happy to swap notes on parsing edge cases anytime.
Congrats on the great launch! Like the skipping bank linking! Receipt emails carry the service name and renewal date while bank statements give you "PAYPAL *XYZ 9.99". Question on Cancel Finder. Cancellation pages get moved and buried on purpose, static index might rot fast etc. Is it a curated list, scraped on a schedule, or resolved live at tap time?
@artstavenka1Β Sharp question β link rot is exactly the failure mode we designed around. It's a hybrid, in three layers:
1. A curated base for well-known services β fastest path, reviewed URLs.
2. Live resolution at tap time for services we don't know yet β we search for the current cancellation page on demand, validate it, and only then open it (no blind redirects).
3. Community feedback as the rot detector: after a cancel attempt the app asks "did this page work?" β downvoted URLs get invalidated and re-resolved, upvoted ones get promoted for the next user.
So a page that moved last week gets healed by the first person who hits the dead end β everyone after them lands on the fresh URL. And you nailed the receipt-vs-statement point: "PAYPAL *XYZ 9.99" tells you nothing; the receipt email has the name, the price AND the renewal date.
The 'out of sight, out of mind' thing hits so close to home. honestly, the biggest barrier with existing tools is definitely the manual entry or the invasive bank syncing, so scanning the inbox without storing the data is a huge UX win.
Curious about how you handle the usage scoring though. if you don't link the bank and don't store the emails, what signals are you using to determine if someone isn't using a subscription? is it based on the frequency of receipt emails, or something else?
@andrasczeizelΒ Good catch β and honest answer: the scoring is deliberately low-tech, because the alternatives (bank data, tracking what apps you use) are exactly the invasiveness we built this to avoid.
Two signals: receipt cadence from your scans confirms a subscription is still alive β you're being charged. Usage comes from you: the app asks "when did you last use this?" and one-tap check-ins update it. No check-in while time passes β the score climbs at 15/31/60-day thresholds, and past a threshold it's flagged as a zombie with a push notification.
So yes, it needs a tiny bit of input from you. That's the trade-off we chose over surveillance β and for the "I haven't opened Netflix since March" cases it works surprisingly well, because those are exactly the ones you never check in on.
Congrats with your launch!
I see more and more products & media getting targeted towards people with ADHD.
As a person with basically zero ADHD myself (I can focus on a single task for hours without any distractions, I have an excellent memory, and enjoy staying on topic), I find it hard to relate to the advertisement, but I'm curious and I do support.
Could you please explain simply what differentiate a products build for people with ADHD from the other products? We all have subscriptions, and many people track them in some way or another.
Which product design decisions were you making following this direction?
Thanks.
Thanks @valzubkov β and genuinely good question, because "for ADHD" is often just marketing paint. For me it's not: I have ADHD myself. I kept forgetting subscriptions existed, and even when I found them, the detective work of tracking them down and the mental load of actually cancelling kept getting postponed forever. I built CentryAI for exactly that: finding the subscriptions I'd forgotten and making cancelling them easy. Here's what that concretely changed in the product:
1. Zero manual entry as a hard rule. Every tracker before this asked you to log subscriptions by hand. If your working memory is the problem, an app that depends on your working memory is dead on arrival. So the inbox scan does the remembering β that single decision is the product.
2. The app assumes you'll forget β it doesn't punish you for it. The zombie score climbs silently while you do nothing, and the app pings you: "you haven't touched this in 60 days, still want it?" It's external memory, not another dashboard you have to remember to open.
3. Friction removal at the action step. Knowing you should cancel and actually cancelling are different tasks when task-initiation is hard. Cancel Finder deep-links to the exact cancellation page β one tap, not a 15-minute hunt that gets postponed forever.
The honest part: none of this is ADHD-exclusive. It's the curb-cut effect β designed for the people who need it most, useful for everyone who's busy or distracted. You with your excellent memory probably don't need the reminders... but you also probably haven't audited your subscriptions in a year π



CentryAI
Thank you @omri_ben_shoham1 β this is exactly the kind of review that helps, because both halves are accurate π Really glad the alert timing landed; "warn me while you can still act" was the whole design goal.
On the App Store miss, two honest notes. First: Apple and Google don't let any third-party app read your store subscription list β it's a deliberate platform restriction, so a direct pull isn't something we (or anyone) can build. Second: Apple does email a receipt for every renewal to your Apple ID address β if that's a different inbox than the one you scanned, that's likely why it slipped through. Scanning the inbox tied to your Apple ID usually catches store-billed subs too.
That said, your suggestion is noted and it's a pain point we take seriously β we're actively improving how the app handles store-billed subscriptions within what the platforms allow. Reviews this thorough are exactly what shapes that work. Thank you π