
Valkyrie
Failed payment recovery for Stripe subscription businesses
Intro
For some context about me, I didn't know how to code (still don't), but wanted to test myself to see if I could build something with AI. I don't mean using lovable or cursor, rather using claude code and a stack that was unknown to me at the time of starting this project.
This isn't my first rodeo at trying to start a business. I've done the drop-shipping thing a few times (didn't work), tried making a travel-connection software (didn't work), designing dog training equipment from scratch (didn't work), custom roasted coffee to order (sorta worked but not really), plus a few other ideas over the last 10 years or so and you get the idea.
Every time I tried something, I would ask for a customer to trust me by way of payment first and then I would deliver the value. This time around I've flipped that on its head with Valkyrie. After reading a few Hormozi sales books I really hitched my wagon first to de-risking my offer and providing value before asking for payment. I can't think of any other offer on the market that does this as well as Valkyrie though I know I am biased.
Process of building Valkyrie
So here is my very summarized process for how I went about building Valkyrie (and this is aided by Claude bc it's taken like 3 months and Claudes memory is better than mine!)
I used Claude as my co-founder for product strategy, architecture, and most decisions along the way about what to build, how to structure the business model, what compliance looked like, how to position it. Then I used Claude Code to actually write the code and Claude Code + v0 by Vercel to design the UI. As mentioned, I'd never built a SaaS product before.
The stack ended up being:
Next.js + Vercel for the app and hosting
Supabase for the database
Stripe Connect so merchants can connect their own Stripe accounts (via OAuth so it only ever reads your payment data, we intentionally designed it to never be able to move money or change setting)
Resend for the recovery email sequences
v0 for UI generation
Claude Code for everything backend
There was a lot of testing as I went (which I think is the right way to go about it) but I wanted to ensure that I had each layer locked down before I added another. To be honest, I probably spend too much time on UI, coloring, light vs. dark theme etc.
Product Status and Where Im at
Valkyrie connects via OAuth to a merchant's Stripe account, classifies every failed payment (what Stripe already recovered on its own vs what's genuinely recoverable), and sends a branded email sequence to their customers asking them to update their card. The merchant only pays 25% of what actually lands back in their account after a 7-day hold. If we recover nothing, they pay nothing. It's specifically de-risked for the merchant and Valkyrie charges an equivalent of a "commission" so to speak once value has been delivered. This also allows merchants to guarantee their return on investment since our charge is dependent the dollar value we're trying to recover.
The thing I'm most proud of is we explicitly separate what Stripe recovered on its own from what we found. Most recovery tools I've seen don't make that distinction, they take credit for everything. We only bill on the lift above Stripe's baseline.
What I'm figuring out now: getting my first real merchant. The product is live at valkyriemrr.com, the infrastructure is solid, and I've done all the compliance work (TOS logged before activation, CAN-SPAM compliant emails, webhook signature verification, the works). But I've never done this kind of outreach before and I'm learning as I go.
Lastly
If you run a SaaS on Stripe and want to see what you're actually losing to failed payments it's free to connect and see your real number. No card required, you only pay if we recover something.
Would genuinely love feedback from anyone who's dealt with this problem, or any advice on what to adjust, add, account for etc. It's been Claude and I up to this point so I welcome feedback!
Cheers,
Bruce
About
I wanted to see if I could build something (as a non-technical person) that offered risk-free (or very limited risk) value. I know through research that companies lose money via failed payments and wanted to solve it!

Comment