Mustafa Aygun

@nicholas Thanks Julian! Stripe integration is already there, that’s the core of the product right now. Shopify isn’t on the roadmap since it’s not a billing/subscription platform itself, but if you mean other billing sources like Paddle or RevenueCat, those are planned for later phases. Prioritizing depth on Stripe first before expanding sources. Appreciate the question and the upvote!

@converter247 Thanks for the support, appreciate it!

@adambadar Both of those are going straight into my notes. The involuntary/voluntary split is the more urgent one, right now everything just says “at risk” with a reason attached, but you’re right that a failed card and a scheduled cancellation need completely different playbooks downstream.

The time-in-risk-state lag is something I’m not tracking yet at all, and now I can’t unsee how useful that would be. That’s basically the difference between “flag” and “was this actually actionable.”

Not building the fix today, but this is exactly the kind of feedback I hoped for by launching before it was perfect. Thank you.

@ksparth12 Thanks Parth! Early feedback has mostly been founders telling me what they already handle themselves, which taught me more than validation would have. Still figuring out who this is most useful for.

@zackashdown Thanks Zack! That’s exactly the blind spot I built this around, billing friction churn is invisible until it’s already cost you money. Appreciate the kind words.

Most SaaS founders find out a customer churned the moment the payment fails. By then it's already too late, the signal was sitting in their Stripe data the whole time, just never surfaced. Loyaltide connects to Stripe read only and shows which customers are at risk right now: failed payments, downgrades, cancellations already scheduled. Every score comes with a plain reason, not a black box number. Built it solo, background is in UX design, so this launch was as much about learning to ship as it was about the product itself. Free 7 day trial, $49/mo after. Would love feedback from anyone running a subscription business on Stripe.

The multi-model comparison angle is smart, translation quality varies so much between models that seeing them side by side probably saves a lot of second-guessing. Curious how the agent decides "most accurate" when the models actually disagree, is that based on some scoring heuristic or just picking a default model per language pair?