About

Plainspoken takes the rough sentence you were going to type into an AI and rewrites it into one that works. Then it tells you what it changed and why, because the goal is that you stop needing it.

It's built for people who keep being told this technology will change everything, and who have so far found it mostly annoying. That's not a you problem. Nobody was handed the instructions.

What happens to what I type?
It's sent to an AI model to be rewritten, and the result comes back to you. This site doesn't save it. We also ask the service in between, OpenRouter, to use only providers that have a zero-data-retention policy — that's a promise those companies make, which we can ask for but can't inspect. There are no accounts, so nothing is tied to your name. Even so — don't paste in passwords, card numbers, or anything you'd mind a stranger reading. That's good practice with any AI tool, not just this one.
Does it cost anything?
No. There's a cap on how many rewrites one person can run in an hour, which keeps the bill survivable. If you hit it, wait a bit.
Will it write the thing for me?
No, and that's deliberate. It writes the instruction. You take that to ChatGPT, Claude, Gemini, or whatever you already use, and that tool does the work. This is the bit that goes before.
Can I trust what the AI tells me afterwards?
Not blindly. A better prompt gets you a better answer, not a guaranteed true one. These tools state wrong things with total confidence. For anything that matters — medical, legal, money — check it against a real source or a real person.
Who made this?
Robert Sweetman built it, working with Claude Code, an AI coding tool from Anthropic. The person decided what it should do and how it should treat you; the AI wrote most of the code. The rewrites themselves come from Claude Haiku, an AI model made by Anthropic, reached through a service called OpenRouter. The code is public at GitHub, so you can check any of this.

How this site works

This site asks you to trust it with your words, so here is how it works, in plain words, including what isn't finished.

Who runs it

Robert Sweetman, on his own. No money, sponsors, or investors are behind it. No ads, no tracking, no analytics.

Services it uses, and what each one sees

  • Vercel hosts the site. It sees each visit, including IP addresses, in short-term logs.
  • This site uses your IP address, in memory only, to cap rewrites at 15an hour. It isn't saved.
  • OpenRouterpasses your words to Claude Haiku, an Anthropic model. We ask OpenRouter to use only providers with a zero-data-retention policy. That's a promise those companies make; we can ask for it but can't inspect it.
  • “Open in ChatGPT” and “Open in Claude” put your finished prompt in the link, so it goes to that company under its own rules.
  • GitHub stores the public code.

What we keep

Nothing. There are no accounts and no database. Your prompt and its rewrite aren't saved, and blanks you fill in stay in your browser.

How it was built

Robert decided what it does and how it treats you. Claude wrote most of the code with him in Claude Code, Anthropic's coding tool, using the newest model available each time: Claude Opus 4.8 (July 2026), Claude Fable 5.1 (September), and Claude Opus 5.5 (October), according to the project's records. The code is MIT-licensed. The lessons and page writing are free to reuse with credit, under CC BY 4.0.

The idea was inspired by Prompt Cowboy. Plainspoken's design and code are its own.

What isn't finished

  • No lawyer, teacher, or other expert has reviewed the lessons yet. Lessons that make legal points link their sources so you can check them.
  • Rewrites come from an AI and can be wrong. Read yours before you use it.
  • The rewriter answers in your language, but the site's own pages are only in English.
  • The hourly cap is counted per server, so it's looser than it sounds.

Tell us what's wrong

Criticism is welcome when it comes with a reason. Open an issue on GitHub and say what's wrong, why it matters, and, if you can, how to fix it. Every fix is recorded in the project's public history.

Want the short version of the skill itself? Nine lessons, a few minutes each.

Why I built this

Over the past few years I noticed most prompting sites were built for developers. But the people I talked to about AI who weren't in tech were mostly intimidated by it — unsure how to make it work for what they actually needed.

AI has improved my quality of life. I built this to help others get there too: openly, for free, and with nothing asked in return. I wanted to take some of the mystery out of it — and, if it's possible, teach you enough that you won't need this site.

— Robert Sweetman · The code is open on GitHub.

Thank you to my parents, Joanne and Robert Sweetman Sr. ❤️