Download Sobes Copilot

Minor fixes and improvements

Minor fixes and improvements

Minor fixes and improvements

Minor fixes and improvements

Minor fixes and improvements

This release adds a new feature specifically for macOS.

Previously, the app was listed in system interfaces under the "technical" name "identityproxy." Now it's literally empty—without a label or visible icon.

Installation also has a new look.

The installer now contains an empty, untitled icon. Simply drag it into Applications as usual.

Less clutter when sharing your screen

If you open the system screen recording indicator, our app will still appear in the list. But thanks to the empty name and icon, it's practically impossible to notice when selecting a browser or a calling app.

This release significantly speeds up screenshot processing. You can now choose what matters most in a given situation: maximum response quality or speed.

Quality vs. Speed

• "Quality" mode is ideal for complex screenshots, diagrams, and interfaces.
• "Speed" mode reduces latency by 25–75%, depending on the selected model.

This setting is available when selecting a model, within quick actions, and for model-specific hotkeys. You can save a preferred mode for each scenario.

Custom Model Rotation

A "Custom List" mode has been added to the settings. Add the models you need and arrange them in your preferred order.

Each time you press the hotkey, the system switches to the next model. After the last model, the rotation starts over from the beginning.

In this release, we've made the desktop client as customizable as possible: we've added full meeting panel customization, a new message order in the chat, and improved handling of long code blocks.

Customize the meeting panel

A new "Meeting Panel" section has been added to the settings.

Now you can drag and drop any buttons between the top panel and the input field, rearrange them, or hide them in "Hidden."

Customization occurs directly in the interactive preview—elements look and are arranged just like in a real session.

If there are no elements left at the top, the panel will automatically disappear and take up no space. And if only one button remains at the bottom, it will be compactly placed next to the send button.

Choose where new messages appear

In the chat settings, you can now choose the message order:

• oldest messages on top;
• New messages at the top.

In the second mode, new and emerging replies remain at the top of the chat. This is especially convenient during a meeting, when you don't want to constantly look down at the bottom of the screen.

Auto-scrolling and switching to new messages work according to the selected order.

Long code blocks are now easier to read

In the chat settings, you can enable a height limit for large code blocks. Long code blocks will be displayed compactly and scroll within the chat, without stretching the entire chat.

If you need to examine the code in more detail, the block can be temporarily expanded to the available panel height and then collapsed back down. The scroll position is maintained.

We've completely redesigned the notes section on both the website and the app – they now look and function identically.

What's been added:

• Bulk actions: you can now not only turn off all notes from a RAG, but also turn them on.
• Visualization of RAG chunks: you can now see how the AI ​​understands notes.

What's been reworked:

• Aligned the interface, corrected awkward margins and other minor issues.
• Made the bulk actions panel more compact and user-friendly.

Minor fixes and improvements

Minor fixes and improvements

Minor fixes and improvements

Minor fixes and improvements

Minor fixes and improvements

⌨️ Added a hotkey for switching models

The hotkey can switch models only within the current provider or cycle through all available models – you can select the desired mode in the settings.

There is no standard combination – you can create a convenient hotkey in the settings.

💼 Added the ability to specify a vacancy for existing meetings

Now you can specify a vacancy not only when creating a meeting, but also add it to an existing one.

After saving, the information will automatically update in the interview tracker, so you don't have to edit the meeting in multiple places.

🧠 Removed the Knowledge Base

Unfortunately, the Knowledge Base didn't live up to expectations.

In practice, it often added unnecessary delays rather than speeding up responses, so we decided not to keep this feature for the sake of adding features and removed it from the app entirely.

Minor fixes and improvements

Minor fixes and improvements

Minor fixes and improvements

We've updated the model selection and added AI reasoning levels.

🧠 Control the depth of your answer

You can now select the reasoning level before generating a response: off, low, medium, high, or maximum.

The higher the level, the longer it takes to generate a response, but the higher its quality.

By default, reasoning is off—quick answers will remain fast.

⏱ Speed ​​​​is immediately visible

The model selection displays the estimated time until the answer starts, separately for each reasoning level.

For simple questions, you can leave reasoning off, and for complex tasks, code, and system design, select high or maximum.

⚡ Works with quick actions and hotkeys

The selected model and reasoning level can be saved for quick actions or assigned to a hotkey.

Minor fixes and improvements