One prompt, different AI tools

What Typo knows, and what it does not

• It knows the app or site. In Auto mode Typo reads the app name, and in browsers the page address with the tab title as a fallback, to decide whether you are in an AI tool.
• It does not know the model. It cannot tell which model is selected, which tokenizer it uses or what your plan costs.
• It does not see the conversation. It only works on what you type, on your computer.

Chat assistants

ChatGPT, Claude, Gemini, Copilot and similar chat tools handle everyday language well. Level 3 to 5 suits most questions. Usage also depends on chat length and attachments, so start a new chat for a new topic.

App builders

Lovable and similar builders turn a short prompt into real work in your project. The size of the change drives the cost, so keep scope and acceptance criteria exact and stay at level 3 or below. From Typo 0.12 lovable.dev is on the Auto list; Ctrl+Alt+T on Windows shortens a finished prompt.

Coding tools

Coding assistants work with files, tests and errors. Typo never changes code, paths or flags, and it stays out of terminals and code editors entirely. For prompts you type in a chat, levels 1 to 3 are the safe choice.

Research

For research, dates, sources and uncertainty matter more than a few words. Put exact source requirements in quotes, and Typo leaves them alone.

A quick guide to levels

• Everyday questions: level 3 to 5.
• Long prompts with background: level 4 to 6.
• Build or code requests: level 1 to 3.
• Short, simple instructions: up to level 8.
• Levels 9 and 10: read the result first, or turn on safe mode.