> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tess.im/llms.txt
> Use this file to discover all available pages before exploring further.
# Preferences
The Preferences section centralizes the settings for personalizing your experience on Tess. Here, you control everything from the interface language and visual theme to advanced settings for chat behavior, memory, subtasks, and file search — all in one place.
### **Where to find it**
1. Log in to your Tess account
2. Click on your profile icon in the bottom-left corner
3. Select "Settings"
4. Click "Preferences"
### **Available settings**
Currently, the Tess platform is available in Portuguese, English, and Spanish. Select the desired language so that all menus and interface texts are translated.
Choose the platform’s visual theme:
| Option | Description |
| :----- | :----------------------------------------------------------------------- |
| Light | White background — default theme |
| Dark | Dark background, ideal for low-light environments |
| System | Automatically adopts the theme configured in
your operating system |
Defines which agent is loaded by default when opening the main Tess AI chat.
* By default, the selected agent is Default (Tess general agent)
* You can switch to any of your own agents created on the platform
* Useful for those who always work with a specialized agent and want to access it directly when opening the platform, without needing to navigate to it manually
Usage example: If you have an agent configured for customer support or data analysis, you can set it as Home Agent and start each session with the right context.
Settings that control the behavior of the chat and Tools.
1. **Operations** Defines the maximum number of operations (requests) that Tools can execute to generate a single response in the chat. Maximum value: 25 operations
* Higher numbers allow more elaborate, multi-step tasks
* Lower numbers limit the scope of actions and reduce resource consumption
This limit applies individually per response, not to the conversation as a whole.
2. **Maximum Subtasks per Prompt** Controls how many subtasks the AI can create internally when processing a single prompt. (Learn more in the article: Link)
3. **Memory Saving Mode** Optimizes memory usage during conversations, reducing context consumption. (Learn more in the article: Link)
4. **Memory Boost** Expands the AI’s context retention capacity during the conversation. (Learn more in the article: Link)
These settings control how Tess processes and searches for information within files attached to conversations. They are technical chunking parameters — the way the file content is divided into pieces for analysis.
Experimental feature: these options are still in testing and may change.
1. **Chunk Size (Fragment Size)**
Defines how much of the file is analyzed at each step when answering a question.
| Value | Behavior |
| :------------------------------------- | :------------------------------------------------------------------------------------ |
| 512 characters
*(More precise)* | Smaller fragments — better for finding specific details
and exact answers |
| 1024 characters
*(Default)* | Balance between precision and context |
| 2048 characters
*(More Context)* | Larger fragments — better for understanding the overall picture
of the document |
**When to use each extreme:**
* Use lower values when you need precise answers about specific excerpts (e.g.: finding a clause in a contract)
* Use higher values when the question requires understanding the broad context of the document (e.g.: summarizing the central idea of a report)
2. **Chunk Overlap (Overlap Between Fragments)**
Defines how much content each fragment shares with the next fragment.
| Value | Behavior |
| :------------------------------------ | :------------------------------------------------------------------------------------ |
| 0 characters
*(Less Overlap)* | No overlap — lighter search, with no repetition |
| 128 characters
*(Default)* | Moderate overlap — recommended balance |
| 960 characters
*(More Overlap)* | High overlap — better when an idea extends across
several parts of the document |
**When to adjust:**
* More overlap: when answers seem "cut off" or lose context between paragraphs
* Less overlap: when answers repeat the same information too much
**Best practices**
* Chunk Size + Chunk Overlap go together: try increasing the overlap if you increase the chunk size, to prevent important information from ending up at the "edges" of fragments without context
* Home Agent: set the initial agent only when you have a consistent workflow. For general use, keep Default
* Operations: if Tools are returning incomplete answers in complex tasks, try increasing this value before reformulating the prompt
### Important notes
* Search Settings settings are experimental and may be changed or removed in future versions
* Changes to preferences are saved automatically and applied immediately
* Chunk Size and Chunk Overlap settings affect only search in attached files — they do not impact responses based on plain text
The Preferences section gives you fine control over how Tess behaves, from appearance to how it processes your files. It is worth exploring each setting — especially Search Settings if you work with documents — to extract more precise results aligned with your workflow.