The Acme assistant is a chat that does real work: it answers questions, drafts and rewrites text, builds plans and checklists, and reads the files and links you give it. It sits at the top of this page — type your request into the box and press Enter. There is nothing to install and no account to create for the first messages.
This page explains what the assistant is good at, how to get a useful answer on the first try, and what to check before you rely on anything it writes.
What the assistant does
The assistant handles four kinds of requests. Most real tasks combine two or three of them in one conversation.
Answers
Ask a question in plain words and get a direct answer, with sources when it searched the web. Follow-up questions keep the context, so “and what about the cheaper option?” works without repeating yourself.
Drafts
Give it a goal and an audience and it writes the first version: an email, a page section, a product description, a reply to a review. Ask for changes the same way you would ask a colleague — “shorter”, “warmer”, “add the price”.
Plans
Describe a job with a deadline and it returns the steps in order, with dates. Plans are easy to edit: move a step, drop one, ask for a version for a smaller team.
Summaries
Paste a long text or attach a file and ask what matters in it. The assistant can also compare two documents and list what changed.
How to get a useful answer
The assistant answers what it is given. “Write an email” gets a generic email. “Write a two-paragraph email to existing customers announcing that we close at 4 pm on Fridays from next month, friendly, no apologies” gets one you can send.
Four things are worth putting in the first message:
- what the result is for and who will read it
- what it must include, and what it must not
- how long it should be
- an example of the tone you want, if you have one
| You write | You get |
|---|---|
| A topic | A general answer you will rewrite |
| A topic and an audience | A draft that is roughly right |
| A goal, an audience, the facts and a length | A draft you can edit in two minutes |
Files, links and languages
Use the paper-clip next to the message box to attach documents and images, or paste a link and ask the assistant to read the page. It answers in the language you write in, and the interface switches between more than forty languages.
What to check before you rely on it
Everything the assistant writes is generated, which means it is sometimes wrong. Check names, numbers, prices and dates against the source before you publish or send anything. For legal, medical or financial decisions, treat the answer as a starting point and ask a qualified person.
More about how conversations, plans and limits work is in the documentation, and the pricing page lists what the free tier covers.
Frequently Asked Questions
Is the assistant free?
Yes. You get a number of free messages every day with no sign-up and no card. Images and the strongest models need a plan.
Do I need an account?
Not for the free messages. Signing in keeps your conversations across devices; without it a conversation stays in the browser where you started it.
Which languages does it speak?
Write in any language and it answers in that language. The interface itself is available in more than forty languages.
Can it read my files?
Yes. Attach documents or images with the paper-clip next to the message box, or paste a link and ask it to read the page.
Is my conversation used to train models?
No. What you share is used to answer you and stays in your account.
What if the answer is wrong?
Say what is wrong and ask again. If it keeps missing, add the facts it needs to the message — the answer is only as good as what it was given.