What is Mantle

Mantle (heymantle.com) is a SaaS product that sits on top of Shopify's App Billing for managing your app's payments and general sales/marketing flow. It's something I've eyeballed off and on between when I was working on Shoplift and our own platform Raptify. It helps manage all of your Shopify payments, your contacts, set up tiers and discounts and usage tracking and all that good stuff, sounds great right? This all takes significant time to set up…

So… Why Didn't We Use It?

Initially we shot it down because of requirements (and we ended up building out before Mantle was even a thing), but with Raptify we were much more aligned: we needed stronger usage-based billing, we wanted to pull all feature sets from billing information, and I had zero sales-pipeline requirements beyond "put money in my bank account". Wonderful, the gaps are not a problem, and the benefits are great… or so I thought.

So we heavily developed with Mantle in mind, wiring up to their tooling, manually loading our plan data, testing, validating. Things were going sort of well minus one major roadblock.

I had no good way to keep our 6 environments synced plan wise (staging + hotfix + production + 3 engineers having their own), this basically meant we were chasing a lot of bugs that boiled down to poor plan configuration in Mantle. Surely we have a way to keep this synced in Mantle right?

Nope, well – what about their API? Nope, couldn't do what we needed.

This missing functionality was going to be a deal-breaker for us, we've already spent a significant amount of QA time chasing bugs that were configuration mis-matches, doing it manually was a huge time sink and would be error prone…

When AI Focus Kills Your Product

So I asked – "How can we keep these in sync? Or have any sort of copy ability? Is this on the roadmap?"

No we don't support that, but maybe you can use our MCP to do that?

What? That had been their response to a lot of people lately, I had noticed there'd been a significant dip in focus on maturing the platform and going all-in on messing with their MCP a lot (at least from my perspective), most of their updates seemed about it, a lot of answers were "just use the MCP" without clarity on if the MCP provided the additional exposure to actually do the things we need, consistently and reliably enough and without additional overhead coaxing the LLM to do what we need…

Should we have just used the MCP? It's a deterministic flow we need – so I'm iffy on that one at best, and the first thing we're seeing get knee-capped in the world of AI is automated flows. Human-in-the-loop at least is probably going to be the last that gets priced to the moon.

Want Me To Use AI That Much? We'll Just Use It To Replace Your Platform

Ultimately, even with Claude, their platform would have been worth the money… if it didn't incur additional overhead to do what we needed.

I'm extremely familiar with many payment platforms, we even built our own back at my OrthoBanc days with Zech and interfaced directly with banks. Stripe being one of my favorite due to a really polished developer experience and a pretty good business ops experience, is an easy choice. I'm familiar with the pitfalls of Shopify's billing system (at least a lot of them), so I draft up a plan and get to work with Claude.

It's maybe a weeks worth of work if even that, and we have exactly what I want: Raptify supports Stripe for standalone accounts and Shopify for in-store customers. One platform, two payment providers, Plan data is internal and seeded for devs and test environments, editable in production, supports all the extra bells and whistles, and I don't need anything else fancy because remember: baseline payment requirements is "I just need to get paid", I'm not going to try to do weird unreliable projections based on Shopify's weird payment behaviors.

So… I Let Out A Sigh Of Relief

Today, we get this announcement: https://docs.heymantle.com/migrating-off-mantle/wind-down – Mantle is shutting down, end of services is in September.

I still had questioned if I made the right call, even with Claude, code costs money to maintain (even without the looming AI cost increases popping up), but with Mantle shutting down, I have no choice but to admit: we made the right choice. Porting off of this would have been a nightmare (especially in 90 days) and I feel sorry for those having to deal with this.

And… There is a Void

On the Shopify Dev Discord TeamDijon pointed out:

This causes a void
Alternatives will surface sooner rather than later

And I'm torn on the viability of filling that void. Why did Mantle decide to shut down? Did they know of some lingering change on Shopify's horizon? Did the business model not pan out to be worth the effort? Were there just greener pastures their team was drawn to? Hard to tell right now…

We'll see, but I don't feel those platforms will really be on our radar.

They're Not The Only Ones

I feel I would have stuck around if they leveraged AI to improve their platform instead of trying to add the latest AI-tooling bells-and-whistles to their product. They're not the only product that I feel ignores the thing that is actively making them money, and the vertical that could continue to make more money, to play with the hottest new fun tech stack (don't worry, we have an article about how Engineers get distracted with fun toys at the cost of your company coming up soon).