Getting Started
Using the Chat Input Bar
Overview
The chat input bar at the bottom of the screen is your control centre: type here, attach here, run a skill here, pick a tool here, start a voice call here.
Get started
1. Type and send
Press Enter to send, or Shift + Enter for a new line. The box grows as you type.
You do not need a tool for most things. Ask for a slide deck in plain words and Torki builds one in the chat — see Making slide decks in chat.
2. Run a skill with /
Type / at the start of an empty message to open the Skills list. Pick one and a Skill: {name} chip appears above the box; the skill applies to the message you send next. See Skills.
3. Open the Tools menu
Click the round + button at the bottom left of the input — its tooltip reads "Tools".
The menu holds:
- Add files or photos — attach a file to your message.
- Take a screenshot — capture part of your screen and attach it.
- Research — runs Deep Research: Torki breaks your question into sub-queries, searches each one, opens the best pages and writes a report. A purple "Deep Research" chip appears above the box and the placeholder becomes "Ask for deep research...". Needs Lite, Arc or Zenith.
- Create image (BETA) — generate an image from your message.
- Canvas (BETA) — build a document, a table or an app in the Canvas editor.
- Guided Learning, Code Analysis, Data Analyst (all BETA) — steer the conversation towards teaching, code review or data work.
There is no Search item — you do not turn web search on. Torki calls the web itself whenever a question needs current information, on every plan, and the pages it used open in a "Sources" panel with a "Sources (N)" pill under the answer that reopens them later.
4. Attach a file
Choose Add files or photos, drag files onto the chat — a dashed "Drop files here" outline appears — or paste an image with Ctrl/Cmd + V. Pasting only works for images; a copied PDF or text file has to be attached or dropped.
Torki accepts PDFs, Word documents, Excel spreadsheets, PowerPoint files, CSV, plain text, Markdown, JSON, ZIP archives, images (PNG, JPEG, GIF, WebP) and video (MP4, MOV, WebM, M4V). The maximum file size is 50MB per file.
Put your question in the same message as the file — "Which region grew fastest?" alongside the spreadsheet — so Torki reads the two together. More in File upload & analysis.
Source-code files are not accepted — .js, .ts, .py, .sh, .php, .rb, .html, .xml and .svg are refused, even though your browser's file picker may offer them. Paste code into the message instead, or upload it to a Project.
5. Start a voice call
Click the Voice button (waveform icon, BETA badge). It only shows when the input box is empty. While a call is running the same button turns your companion's colour and reads "Listening", "Thinking" or "Speaking"; click it again to hang up.
The first time, your browser asks for permission to use your microphone — allow it and the call starts. There is no setup wizard. Choose which companion answers in Settings → Voice.
If you deny the microphone, the button sits on "Stop" without connecting. Reload the page and allow it.
Good to know
A tool applies to your next message only — the chip above the box clears as soon as you send.
The × on that chip does not switch the tool off; it offers to start a new chat. If you just want to cancel a tool, send the message or start a new chat.
FAQ
Where is the paperclip?
Inside the Tools menu, on the "Add files or photos" row. There is no paperclip on the bar itself.
Why did my code file not attach?
Source files are refused for security. Paste the code into the message, or put the file in a Project.
Can I keep a tool on for a whole conversation?
No. Pick it again for each message that needs it.
Is there a tool for slide decks?
No — just ask. "Make me an 8-slide deck about…" in an ordinary message builds a deck in the chat. With Research or Canvas picked, Torki does not build one.
Why has the Voice button disappeared?
It hides while there is text in the box or while an answer is streaming.