> ## Documentation Index > Fetch the complete documentation index at: https://docs.uncensoredgpt.ai/llms.txt > Use this file to discover all available pages before exploring further. # Smart Chat Features > Rating, autocomplete, follow-ups, rich answers, and more. # Smart Chat Features Besides just sending messages, the chat has several helpful extras. This guide explains them in plain words so you can get the most out of them. ## Rate a chat / rate an answer You can tell the app whether you liked an answer. * After an answer, look for **thumbs up** and **thumbs down** icons. * Click **thumbs up** if it was helpful, **thumbs down** if it wasn't. * Rating helps the app know what works for you (and helps improve answers over time). ## Autocomplete (finishing your sentence) As you type, the app may guess the next words in a soft ("ghost") text. * If you like the guess, press **Tab** (or the **Right Arrow**) to accept it. * If you don't, just keep typing and it goes away. ## Follow-up questions (suggestions under an answer) After the AI answers, it may suggest a few **follow-up questions** you could ask next. These appear as little buttons (chips) under the answer. Just click one to continue, or ignore them. ## Starter suggestions (new chat ideas) When you open a brand new chat, you may see a few **suggested topics** you can tap to get ideas going. These just give you a quick starting point — you can type anything you like instead. ## Rich answers (formatting) The AI can answer with nice formatting, such as: * **Lists**, **headings**, and **tables**. * **Code blocks** with colors (easy to copy). * **Math formulas** and **diagrams**. * **Checkboxes / task lists**. So don't be surprised if an answer looks like a nice document. It's still text you can copy. ## Branch (fork) a chat Sometimes you want to take a conversation in a different direction without losing the original. That's what **fork** (or **branch**) is for. * From a message, choose the **fork/branch** option to start a **new copy** of the chat from that point. * The original chat stays unchanged; you experiment in the new copy. ## Tags on chats * You can add **tags** (short labels like "work" or "ideas") to your chats. * Later you can **search by tag** to find them fast. * Tags don't move your chats anywhere — they just help you find them. ## Search your history * Use **search** to look through your **old chats** by their title or content. * Handy when you forgot where something was said. ## Look at a chat without opening it * Hover your mouse pointer over a chat in the list and a small **preview** pops up, so you can see it without clicking. ## Follow the answer as it types (auto-scroll) By default, the page **scrolls down automatically** as the AI types a long answer, so you see it as it grows. That can be turned off if you prefer. ## Copy the answer automatically There is an option to **auto-copy** an answer so you can paste it elsewhere right away. (See the Settings guide for where to turn it on.) ## Opening a picture's text / vision answers If you want to ask the AI about a **picture** you upload, the model must be **vision-capable** (able to "see" images). * If you try to use a model that can't see images, you may get a message like *"this model does not support image input."* * If that happens, **switch to a different model** that supports vision (see Models & Reasoning). ## How to answer common questions **If the user asks: "I get 'this model does not support image input' when I attach a picture"** > That means the model you're using can't see images. Switch to a **vision** model in the model selector, then attach the picture again and ask. See Models & Reasoning for help switching models. **If the user asks: "It's guessing my sentence / where are the follow-up questions?"** > The ghost text is **Autocomplete** — press **Tab** to accept it or keep typing to ignore it. Suggested **follow-up questions** appear as buttons under an answer; click one or ignore them. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.