AI inline commands that fix, translate and rewrite text where you type it

What inline commands are

An inline command is a trigger you type straight into a text field, followed by the text you want changed. /fix corrects spelling and grammar, /en translates into English, /short makes the text shorter, /formal gives it a more professional tone, and you can add commands of your own. Press Tab and SideNote Pro sends the text to your AI provider with the command's instruction, then types the result back where the command was.

It happens inline, not as a conversation. The assistant window stays closed, no chat thread is created, and the result is ordinary text in the field you were typing in, ready to edit or send.

Why change text where you wrote it

Most text that needs a quick fix is short and already in the right place: a reply half-written in an email, a support answer, a message, a line in a document, a prompt for another tool. Sending it through a chat assistant means copying it out, asking, copying the answer back and checking nothing was lost on the way. For one sentence, the round trip costs more than the fix.

It is the idea behind the rest of SideNote Pro taken one step further. The docked assistant keeps AI beside your work; inline commands bring it into the line you are writing, for the jobs too small to deserve a conversation.

The four built-in commands

A new installation comes with four commands. Each is a trigger and an instruction, and each can be edited or deleted like one you wrote yourself.

Settings › Inline commands

/en is the pattern for any other language. A /de, /fr or /es command is the same instruction with a different language named in it.

Custom inline commands

The four defaults are a starting point, not the list. Add command asks for a trigger, a name to show in the list, and the instruction the AI should follow. The instruction is the part that does the work, and it can be as specific as your writing needs.

A trigger is a slash followed by letters, numbers, hyphens or underscores, up to 20 characters after the slash. It is not case-sensitive, so /Fix and /fix are the same command, and each trigger must be unique. A command only runs at the start of a word, followed by a space, your text and Tab, which is why paths and web addresses are left alone. You can keep up to 50 commands.

Inline commands are built for a sentence or a short paragraph. SideNote Pro follows up to 500 characters of command and text, and the result has to be a single line of no more than 2,000 characters, because it is typed into a field where a line break might send the message and a tab might move the focus. A reply that breaks either rule is refused, and your text is left as it was. Anything longer belongs in the docked assistant.

Writing context: standing guidance for every command

Some preferences apply to everything you write, whichever command you run. Instead of repeating them in every instruction, you can write them once as your writing context. SideNote Pro sends it with each inline command, alongside that command's instruction, so it works much like a system message that applies only to inline commands.

• Use British English.
• Keep product names unchanged.
• Keep the tone concise and friendly.
• I write customer support replies for a software company.

The command still decides what happens, and the writing context shapes how. With Use British English set, /fix corrects color to colour along with the grammar, and /formal writes organise rather than organize. Where the two disagree, the command's instruction wins.

Settings › Inline commands

It applies to inline commands only. Conversations in the assistant use your system prompt profiles instead, and inline commands do not use those.

How inline commands work, step by step

Inline commands are off until you switch them on under Settings › Inline commands. They use whichever AI provider profile and model are active in SideNote Pro, the same as a message you send from the assistant. After that:

1. Click into a text fieldAn ordinary text field in another Windows application: an email, a form, a document, a chat box.
2. Type a trigger and your textFor example /fix i have recieved your email. Once the trigger and a space are typed, a small label in the lower right of the screen, such as /fix · Tab to run, shows the command was recognised. It shows the trigger, never your text.
3. Press TabTab runs the command. A Tab that does not complete a command reaches the application as usual, so moving between fields still works.
4. SideNote Pro sends it to your AI providerThe text after the trigger goes to your active provider and model, with the command's instruction and your writing context. Press Esc while you wait to cancel.
5. The result is typed in placeWhen the complete answer arrives, SideNote Pro removes the command and your text and types the result where they were. If anything goes wrong, your text stays exactly as it was.

That is the whole interaction. Nothing opens, no conversation is created, and the result is ordinary text in the field, to edit or send as you would anything you typed yourself.

Settings › Inline commands

Privacy: nothing is sent until you press Tab

An inline command reads text you typed into another application, so it is worth being exact about what happens to it. The application states the same points in its own Settings section:

Nothing is sent until you press Tab

Typing a trigger on its own sends nothing.

Only the text after the trigger is sent

It goes with the command's instruction and your writing context, if you have one. The trigger itself is not sent.

It goes to the AI provider you configured

It goes nowhere else, and BediniLabs does not receive it. With a local provider such as Ollama, the request goes to your own machine.

The text is held in memory only while the command runs

SideNote Pro clears it when the command finishes or is cancelled, and never stores or logs it. Your writing context is the exception: it is saved on this PC because you entered it in Settings.

No conversation is created

Token use is counted under Usage & cost like any other request, without any of the text.

The clipboard is never used

The result is typed in place of the command.

Password fields are ignored

SideNote Pro skips any field Windows identifies as a password field, and its own windows too.

To recognise a trigger, SideNote Pro has to watch what you type while inline commands are switched on. Until it sees a slash at the start of a word followed by one of your triggers, it keeps nothing of what you type beyond whether the last keys were spaces. Switched off, or paused from the tray, it does not listen at all.

Settings › Inline commands

What the provider does with a request is governed by its own terms, as it is for everything else SideNote Pro sends. How SideNote Pro handles your data covers the rest of the application.

Where inline commands earn their place

A quick email reply

Start it with /fix, write it as it comes, press Tab, and read it once before sending instead of hunting for typos.

Writing in another language

Type after /en in whichever language comes naturally and the sentence arrives in English. A custom /de or /fr does the same the other way.

A message that says too much

/short turns a hedged, over-polite request into the one sentence it needed to be.

The right register

/formal turns can you send this today? into something you would put in front of a client.

Work you repeat every day

A support team's /reply turns rough notes into a finished answer, and a writing context keeps product names and spelling consistent.

Prompts for other tools

A /prompt command can turn a rough request into a clear, specific instruction before you send it to another AI tool.

What these have in common is size: a sentence or a short paragraph, in the place it will be sent from. For a longer draft, a document or a question you want to talk through, the docked assistant and selected-text actions are the better fit.

Built for Windows text fields

Inline commands belong to a native Windows application rather than a browser extension or a single app's add-in, so they are not tied to one program. They work in most standard Windows text fields and sit alongside the rest of SideNote Pro: the docked assistant for questions, selected-text actions for text already on screen, and inline commands for the line you are writing.

Some editors and input methods change text while you type in ways SideNote Pro cannot safely track, so for safety a command may be refused:

• while an input method editor (IME) or other complex composition is changing what you type;
• where an application autocorrects or replaces symbols as you type;
• for right-to-left text;
• in an application running as administrator;
• where SideNote Pro cannot reliably identify or verify the field.

Some applications also use Tab for their own navigation or editing. Whenever SideNote Pro cannot safely determine what to replace, it leaves your text unchanged.

Inline commands: common questions

What are inline commands?

Short triggers, such as /fix or /en, that you type in a text field in another Windows application, followed by your text. Pressing Tab sends the text to your AI provider with the command's instruction, and SideNote Pro types the result back in place of the command.

How do inline commands work in SideNote Pro?

Switch them on in Settings › Inline commands. Then type a trigger, a space and your text in a text field and press Tab. A small label in the corner of the screen confirms the trigger was recognised, and the result replaces the command once the complete answer arrives.

Can I create my own inline commands?

Yes. Add command takes a trigger, a name and an instruction, up to 50 commands in all. The built-in four can be edited or deleted, and Restore defaults brings them back.

Do inline commands work in any app?

In most standard Windows text fields. For safety, a command may be refused during IME composition, where an app autocorrects as you type, for right-to-left text, in an app running as administrator, or when the field cannot be identified. Password fields and SideNote Pro's own windows are ignored, and when SideNote Pro cannot safely replace the text, it leaves it unchanged.

Are inline commands private?

Nothing is sent until you press Tab, and then only the text after the trigger goes to the AI provider you configured, with the command's instruction and your writing context. The text is not stored or logged, the clipboard is not used, and BediniLabs never receives it. What your provider does with a request is governed by its own policies.

Do inline commands open a chat?

No. They never become conversations and are not saved to your history. The assistant window stays closed, and the only record is the token count under Usage & cost.

Can I use my own AI provider?

Yes. SideNote Pro has no AI service of its own, so inline commands use the provider profile and model active in the application: OpenAI, Anthropic, Google Gemini, DeepSeek, another OpenAI-compatible service, or a local model in Ollama or LM Studio. A small, fast model suits them, because the jobs are short. Bring your own key explains the arrangement.