Who's behind Stax.
I'm Collier. I'm building Stax alone, out of Gainesville, Georgia, because I got tired of not knowing what my own AI bill was.
Last year I was running a small team and paying for Cursor, Copilot, ChatGPT Team, the OpenAI API, and an Anthropic key. Every month the numbers got bigger and every month I had less idea where they were going. The provider dashboards each tell their own slice of the story. None of them stitch together. None of them care whether you'd rather have the answer in dollars than in tokens.
I looked at the existing observability tools. They're built for engineers who want to debug prompts — not for the person who signs the invoice. They trace API calls and pretend the $840/yr/dev in subscriptions doesn't exist. They charge variable rates that climb the exact moment your usage does, which is the worst possible time for your cost-control software to also get more expensive.
So I started building the thing I wanted. One dashboard. API and subs in the same view. Dollars, not tokens. Flat $49 a month, forever.
The bet behind flat pricing.
Variable pricing on cost-tracking software is a tax on the people who need it most. If your AI bill triples next month, you should not also be paying triple to find out. So Stax is a flat $49 regardless of how many providers you connect or how much you spend. If that ever stops working as a business, I'll figure something else out — but I'm not adding usage-based pricing through the back door. That's the promise.
Where this is going.
Pre-launch right now. The waitlist just opened and I'm reading every pain-point note that comes in. New providers ship every two weeks based on what people actually ask for. If you want a say in what gets built first, the pain-point field on the waitlist form is the most direct line to me.
Idea to live site, four months.
No accelerator, no funding round, no team. Just the steps that got from "I'm annoyed at my AI bill" to a working waitlist.
Four weeks reading Reddit, not coding.
Read every thread I could find about founders and CTOs complaining about AI cost. ~300 posts across 7 subreddits. The same three problems kept repeating — that became the brief.
Provider matrix v1.
Mapped which providers expose billing endpoints, which require OAuth, which need invoice scraping. 8 providers shipped in the v1 matrix. Stripe and Brex feeds confirmed as a fallback for the long tail.
Design system locked.
Built the paper-and-ink design system end-to-end — colors, type, the dashboard kit — before writing a line of product code. The whole site and the eventual product run on the same tokens. No drift later.
Marketing site, waitlist, demo dashboard live.
You're reading the result. Waitlist open, founding price locked in for the first 100 sign-ups, product build now starts.
What I'm not building.
A solo founder with one feature ships. A solo founder with five features doesn't. So here's the explicit no-list.
Things Stax will not become.
- ✕Not a prompt observability tool. The market has good ones. Stax doesn't trace requests, doesn't time spans, doesn't store completions. The bill is the product.
- ✕Not a model-evaluation harness. Picking the right model is your call. Stax tells you what each one cost you last month.
- ✕Not a per-token optimizer. Stax surfaces spend by team, project, and person. It does not rewrite your prompts or route between models.
- ✕Not a vector DB, a RAG framework, or an agent runtime. Anything model-side. Stax sits one layer up — at the invoice, not at the inference.
What's shipped, what's building, what I'm sitting on.
Plain version of the kanban I work from. Updated on every waitlist email I send.