About

Why I built Stormfire

Founder's note — June 2026

Why Stormfire — 创始人故事页 (Why Stormfire / About Page)

目标 URL: stormfire.io/about 或 stormfire.io/why-stormfire 目的: 建立信任，回答"你们是谁，为什么我应该信你们" 风格: 第一人称，诚实，不卖惨，不浮夸 长度: ~1800 字（移动端可读完）

Why Stormfire

The story you'd tell over a beer

In April 2026, I lost my third OpenAI account in two weeks.

Different credit card, different VPN, different billing address. Same result: "Your account has been suspended due to billing risk." No appeal, no human review, no actual explanation.

I had been a paying OpenAI customer for over a year. Built two production apps on their API. Never violated their terms of service. The only thing that changed was that I'd started traveling — bouncing between Tokyo, Seoul, and Bangkok — and Stripe's risk model decided I was a fraud risk.

I tried everything. A friend's US address. A Wise card. A residential proxy in Texas. Each fix worked for about a week, then the next account got flagged.

After my third suspension, I stopped trying to fix the problem on my end. The problem wasn't me. The problem was that the AI API ecosystem had built itself on top of one payment rail (Stripe) optimized for one customer profile (US-resident, US-billed, predictable usage) and treated every other profile as friction.

So I built the bypass.

What Stormfire actually does

Stormfire is an API gateway. You point your OpenAI SDK at us instead of api.openai.com. We forward your requests to the actual model provider, pay them in USD via traditional rails, and bill you back in USDT cryptocurrency.

From the AI provider's perspective, all traffic comes from one entity (us) with one billing relationship. No region triggers, no fraud signals, no Stripe risk score per customer.

From your perspective, you wrote your code once and it just works. Same SDK, same patterns, same models — paid in USDT instead of credit card.

That's the entire product. No magic. No model fine-tuning. No proprietary inference. Just removing the Stripe gatekeeper from the customer-facing side.

Who I'm building for

Stormfire is not for everyone, and pretending it is would be a disservice.

We are explicitly for:

  1. Developers in 30+ countries where Stripe is unreliable. India, Brazil, Argentina, Russia, Indonesia, Nigeria, Vietnam, Turkey, and more. If you've gotten more than one card declined trying to subscribe to a Western SaaS, you know what I mean.

  2. Privacy-conscious builders who don't want to give a card on file to every AI vendor they touch. We accept USDT. You can use a fresh wallet per project if you're paranoid.

  3. Crypto-native indie hackers who hold USDT for trading and treat AI tokens like another consumable. The accounting model fits the workflow.

  4. Builders serving emerging markets whose own users can't pay in cards. The AI costs flow naturally through the same crypto rail you might be using for revenue.

We are explicitly not for:

The honest version: we're a niche product. We're never going to be the AI gateway for everyone. We're going to be the AI gateway for the customer the bigger platforms structurally can't serve.

What I learned building this

Lesson 1: The market for "AI access for the underserved" is bigger than I expected.

In 30 days post-launch, we hit 47 paying customers and ~$1.4k MRR. The conversion rate from sign-up to first top-up is 28%, which is much higher than I expected. The reason isn't marketing skill — it's that the customers we serve don't have alternatives.

When you build for someone who's been told "no" by every other vendor, they don't shop around when they find a "yes."

Lesson 2: Crypto operators have earned their bad reputation, and we need to earn back trust.

I've been ripped off by enough fly-by-night crypto services to know how it feels. The path forward isn't denying that pattern exists — it's being unusually transparent about who I am, what I'm doing, and what could go wrong.

You'll notice we don't have a flashy "Trusted by 10,000 customers" banner. We have 47 customers as of writing this, and I think that's the right number to be honest about.

Lesson 3: The Chinese AI model market is structurally interesting.

DeepSeek, GLM, Qwen, Kimi — these are competitive with Western frontier models on most workloads, and they're 5-10x cheaper. The catch is they're priced for the Chinese domestic market.

We're working on routing high-volume customers to these models when their use case allows, and capturing the margin spread to fund our 0% markup on Western models. This is part of the path to sustainable economics.

What I'm scared of

Three things keep me up at night.

One: I'm too small to offer the trust signals customers should demand.

A real bank has audited reserves. A real custody provider has insurance. I have... my word and a Tokyo VPS. If you're considering topping up a large amount, please don't. Top up small, use it, top up again. We need to earn the right to hold larger balances and we haven't yet.

Two: Crypto regulation could change abruptly.

Japan's crypto framework is relatively mature, but global crypto regulation is in flux. If a major jurisdiction tightens, we'd need to adapt fast or wind down. I have a 30-day notice plan in our ToS, but no guarantees beyond that.

Three: I'm one person on engineering with one part-time helper on support.

If I burn out, get sick, or have a family emergency, the lights stay on (infrastructure is automated) but new features stop and support degrades. I'm working on resilience — hiring a part-time support lead is the next step at $5k MRR.

These aren't reasons not to use us. They're reasons to use us with the right expectations.

What I won't do

A few promises that are easier to make than to keep, but I'll try:

I won't pivot to "Stormfire Premium" with hidden markups. If our path to sustainability requires charging more, I'll raise prices transparently and grandfather existing customers as long as I can.

I won't sell user data or AI prompts. We don't train on your requests. We don't have a side-business in selling aggregated query data. If we ever need to, I'll say so before doing it.

I won't expand into adjacent products that betray the core promise. No NFT integrations, no "Stormfire Token," no DAO governance theater. We're an API gateway. We'll stay an API gateway.

I won't ghost users. If you email us and we don't respond within 48 hours during business days, something is broken. Re-email. Escalate to founder@stormfire.io. We will respond.

What I want to build next

Roadmap, in rough order of priority:

  1. Russian and Spanish language documentation — these are our two highest-friction markets and our docs are English-only
  2. Auto-refill via Telegram bot — for users who don't want to manually top up
  3. Multi-region edge proxies — currently Tokyo-only, want to add Frankfurt and Virginia
  4. More Chinese model support — DeepSeek v4 is in, GLM-5 and Kimi 2 coming
  5. Observability dashboard — token-level usage analytics for teams
  6. Better trust tooling — eventually a multi-sig audited reserves system, but that's a longer-term thing

If you want to vote on priorities, I read every email. founder@stormfire.io.

The honest pitch

Stormfire is small. We launched 30 days ago. We have 47 paying customers and ~$1.4k MRR. Our infrastructure costs $100/month.

We are not a unicorn. We are not "the future of AI." We are a single-purpose tool built for a specific kind of developer who has been failed by the existing ecosystem.

If that's you, $15 USDT is enough to try us: https://stormfire.io/signup

If it's not you, that's fine. The internet has plenty of AI gateways. Use the one that fits your situation.

Either way, thanks for reading this far.

— Founder stormfire.io

Last updated: 2026-06-22 Founder availability: I read every email. Replies within 48 hours, often within 6.