About — why we build tools that never see your data
Small browser-side tools for people who work with language models
We are three people who got tired of the same three chores, and decided to solve them once.
What we build and why
Anyone who uses language models daily ends up doing the same small jobs over and over: guessing whether a prompt is going to cost ten cents or two dollars, rewriting the same instructions for the fifth time, and fighting with a reply that is 95% JSON and 5% apology.
Those jobs are small enough that nobody wants to install anything, and annoying enough that doing them by hand every day adds up. So we built a set of single-purpose tools that open in a tab and close when you are done.
The decision that shapes everything else is that they run entirely in the browser. A tool that helps you handle a client's contract should not require you to send that contract to our server — and once the tool is client-side, there is no server to send it to.
What we optimise for
Nothing leaves the tab
Every tool is client-side. There is no backend that could log your text, because there is no backend.
One job, done well
No dashboards, no onboarding flows, no account wall. Open the tool, get the answer, close the tab.
Honest about limits
The token count is a heuristic and we say so in the interface instead of implying precision we do not have.
How we got here
- 2024 Q3
First internal script
A throwaway token counter written to settle an argument about prompt cost.
- 2025 Q1
Opened to the public
Five tools, a single page, and a strong opinion about not adding a sign-up form.
- 2025 Q3
Model pricing tracker
Manual price edits replaced by a review pass whenever a major lab ships a new model.
- 2026 Q1
Team libraries
The first paid feature, added only because teams kept asking to share prompt templates.
At a glance
- 10
- tools shipped
- 0
- bytes of your text stored on our servers
- 5
- model families priced
- 3
- people on the team
The team
Chen Yiyu
Founder & frontend engineer
Spent six years on developer tooling and has strong feelings about bundle size.
Wang Siyuan
Technical writer
Writes the documentation and, by extension, most of the copy you are reading.
Li Jing
Data analyst
Keeps the model pricing table honest and reviews every estimate heuristic.
Try the toolkit
No account, no install. Start with the tool most people start with.