AI writing assistant

It is an optional helper in the editor that can draft text, improve, shorten, expand or re-tone what you write, and suggest titles and poll or quiz choices. It runs only when you press one of its buttons.

In the editor's toolbar, press the AI button ("Write or improve with AI"). It is in the text editor for posts, questions, polls, MCQs, answers, services and jobs, but not in comment and reply boxes. You need to be signed in.

It has five quick actions: Improve, Fix grammar, Shorten, Expand and Friendlier. You can also type your own instruction, for example "write a short post about the street fair".

Yes. If you have not written anything yet, type an instruction and press Generate, and it drafts the text. Read and edit the draft before you publish.

If you select some text before opening the assistant, it works on your selection. If nothing is selected, it works on the whole text.

Yes. The result shows as a preview first. Press "Use this" to put it in, "Regenerate" for another version, or "Cancel" to keep your text as it was.

Yes. Every signed-in member can use it at no cost, on the free plan too, up to an hourly usage limit.

The assistant has a Model menu with several models, such as OpenAI gpt-oss-20b (the recommended default), NVIDIA Nemotron, Google Gemma and "Auto (best free)". Your browser remembers the model you picked.

It lets the AI provider pick whichever free model is available at that moment. Choose it if the model you picked is busy or keeps failing.

"Could not respond" means the model did not answer this time: try again, or pick another model such as "Auto (best free)". "Not available right now" means the assistant is switched off for the moment; try again later.

The assistant takes at most 8,000 characters at a time, counting your text and your instruction together. Select a part of your text and run it on that part.

Each member can use the assistant a limited number of times an hour, and there may be an overall limit too. Wait a while and try again.

Your text leaves Ur Next Door: up to 8,000 characters of your draft, your instruction and the model you chose are sent to our AI provider, OpenRouter, and on to the company behind the model you picked. It is sent only when you press a button in the assistant.

See the Privacy Policy for details.

No. Ur Next Door does not store your prompts; it keeps only a per-member counter to enforce the usage limit.

It can. Depending on the provider and tier in use, your input may be kept or used by that company to improve its own models, under its own privacy policy.

Yes. Do not put confidential or sensitive information, or personal information about other people, into the assistant, because your text is sent to an outside AI company.

No. AI output can be wrong or can look like existing material, and Ur Next Door does not promise that it is accurate or original. Check facts before you publish.

You are. You are responsible for what you publish, whether or not the assistant helped write it, so review it before posting.

Yes. It replies in the language you write in, and you can ask it to translate, for example "translate to Hindi".

Yes. In the create form, "Suggest a title" drafts a title from your text, and "Suggest choices" fills in the options for a poll or MCQ from your question.

You need to be signed in to use the assistant. It is also hidden whenever the assistant is switched off on Ur Next Door's side. Comment and reply boxes do not have it; it is in the bigger text editors, such as the one for a post.

No. The assistant only runs when you press one of its buttons, and text from the assistant window goes into your editor only when you press "Use this". "Suggest a title" and "Suggest choices" fill in their boxes, which you can still change. You still publish yourself.

Not yet. The assistant cannot fix your SEO score or its tips for you. You can follow the tips shown under the form and ask the assistant to improve, expand or restructure your text yourself.