[New Add-on] Vision Assistant Pro: Your Interactive AI Copilot for NVDA (Powered by Gemini)

not connected? There is no such message in the plugin! Please check again.

Hello. Bulgarian has been added to the new version, 2.6. Enjoy.

Hello. That's not the case! Unless you are translating in the browser! To do this, place the text in the clipboard and press 'y' instead of 't'.

Dear Users,

Please download and install the latest build of the add-on from the project's GitHub page, which is linked in the original post. It appears that a significant portion of users are still running outdated versions.

Due to time constraints, my ability to visit this thread frequently to respond to comments is limited. For any issues or suggestions, please raise them directly in the Issues section of the project's GitHub repository.

Best Regards,

Mahmood

Hello everyone. The download link has been updated to version 3.1.0. Enjoy.

Hello, yes, this issue exists and will be fixed soon in version 3.5.0.

Thank you so much for your kind words and feedback, Maurizio! I'm glad to hear the MP3 fix and video descriptions are working well for you.

Regarding your suggestions, I'll be adding support for X (Twitter) very soon. As for local video support, I am currently thinking about it and evaluating the best way to implement it.

Regarding the privacy concerns you mentioned, I want to reassure you that since this addon is open source, everything is transparent and there is no need to worry at all. Your trust and support mean a lot to me—thank you again!

Hello, thank you for checking on me. I am well, but internet access has become very difficult, and it's unclear if I will be able to access it in the future. I will try my best. I have never worked with TikTok before. Please send a link to a video. If I can and my internet is stable, I will implement it.

You can get the latest version from the original post.

Hello everyone. The new version of the addon has been released. Support for TikTok videos has also been added!

Hello everyone.
The new version has been released.
You can use it.

Hello everyone, I'm glad the addon has been useful to you. I try to always add more features to the addon. Regarding donations, unfortunately, I am in Iran and do not have access to PayPal. If you would like to support the project and do not have the option to support with cryptocurrency, you can purchase an Apple USA gift card and send it to me via private message. Thank you in advance for your support.

hi. I'm glad the addon has been useful to you. I try to always add more features to the addon. Regarding donations, unfortunately, I am in Iran and do not have access to PayPal. If you would like to support the project and do not have the option to support with cryptocurrency, you can purchase an Apple USA gift card and send it to me via private message. Thank you in advance for your support.

Hello to you, I'm glad the addon has been useful for you.
1. I am blind, but I don't think this section relates to the display, because it uses the default NVDA settings.
2. This also seems to use NVDA settings.
3. The choice is entirely yours. Check it out.
4. I think it's about 10 minutes.
5. You can assign any key to any section in Input Gestures.
Regarding donate, if you would like to support, and if you cannot use crypto, you can purchase an American Apple gift card and send it to me privately on Applevis.

Hello everyone,

Vision Assistant v4.5 is now officially available! This update brings several significant improvements, including the new Advanced Prompt Manager, full proxy support for all API requests, and an enhanced Document Reader experience.

Download Instructions: Please always use the official download link provided in the original post at the top of this thread to ensure you are getting the latest verified version.

Support the Project: If you find this addon useful and would like to support its continued development, please consider making a donation. Due to international banking restrictions in my region, I am unable to use traditional platforms like PayPal. To address this, I have updated the "Donate" dialog within the addon settings with alternative ways to contribute, including Apple Gift Cards (US region) and Cryptocurrency.

Your support is greatly appreciated and helps keep the project evolving. Thank you all for your feedback and for being part of this journey!

Thank you so much for your incredibly kind words! Your feedback truly made my day and is exactly the kind of encouragement that fuels my passion for this project.

It's fantastic to hear that the plugin has been able to consolidate the functionality of multiple separate tools, like translation and OCR, into one seamless experience for you. That was precisely one of the core goals, and knowing it's making such a significant difference in your daily workflow is incredibly rewarding.

And your comparison to Be My Eyes for PC, especially given its previous reputation, is incredibly high praise. This kind of detailed and enthusiastic feedback is invaluable; it not only validates the hard work and countless hours put into development but also inspires me to continue improving and expanding its capabilities.

If you feel inclined to support the ongoing development of this project and my motivation to keep bringing new features and improvements, a donation would be greatly appreciated. Every contribution, no matter how small, helps sustain the effort and passion behind it.

Thank you once again for your wonderful words and for being such an enthusiastic user!

Hello. When you update an addon through NVDA, there is no difference compared to updating it through the addon itself. The update methods for both NVDA and addons are the same. However, it's strange that you receive an error, whereas you should be getting the message "You have the latest version".

Hello everyone,
Vision Assistant v4.6 is now officially available!
Download Instructions: Please always use the official download link provided in the original post at the top of this thread to ensure you are getting the latest verified version.
Support the Project: If you find this addon useful and would like to support its continued development, please consider making a donation. Due to international banking restrictions in my region, I am unable to use traditional platforms like PayPal. To address this, I have updated the "Donate" dialog within the addon settings with alternative ways to contribute, including Apple Gift Cards (US region) and Cryptocurrency.
Your support is greatly appreciated and helps keep the project evolving. Thank you all for your feedback and for being part of this journey!

Hi everyone,
I’m happy to share that Version 5.0 of Vision Assistant is now live!
The highlight of this major update is the new Multi-Provider architecture, allowing you to switch between Gemini, OpenAI, Mistral, and Groq. Most importantly, it now supports Custom API URLs, giving you full control over your endpoints.
This version was developed under very difficult conditions, mostly locally and offline. Please refer to the main post of this thread for the download link and full details.
On a side note, since I am completely cut off from standard banking like PayPal or Stripe due to international restrictions, I’ve added Apple Gift Cards (US Region) as a donation bridge. For many in this community, this is the most accessible way to show appreciation for the thousands of hours put into this original work. You can send them directly to: [email protected].
Thank you for your continued support!

Changes for 5.5 (The Automation Update)

• AI Operator (Autonomous Control - Shift+A): This is the crown jewel of v5.5. Vision Assistant Pro has graduated from being a passive assistant to becoming your personal AI Operator. It doesn't just describe the screen—it takes command.
  • How it works: You can now give verbal instructions to operate your PC. For example, in a completely inaccessible application where your screen reader stays silent, you can press Shift+A and type: "Click on the Settings button" or "Find the search field, type 'Latest News' and press enter." The AI visually identifies the elements, moves the mouse, and executes the task for you.
  • Performance Note: This feature is optimized for Gemini 3.0 Flash (Preview), delivering incredibly fast and intelligent responses that can handle even the most complex UI layouts.
  • ⚠️ API Usage Warning: Because the AI Operator needs to "see" exactly what's happening to be accurate, it sends a high-resolution screenshot with every step. Please note that frequent use will consume your API quota much faster than standard text-based tasks.
• Visual UI Explorer (E): Tired of navigating through "unlabeled buttons"? Press E to activate the UI Explorer. The AI will scan the entire window and generate a list of every clickable element it sees—including icons, graphics, and menus. Simply pick an item from the list, and the AI Operator will click it for you. It’s like having an "accessible layer" on top of any app.
• Context-Aware Smart File Action (F): The "F" key has been completely overhauled. It no longer assumes you only want OCR. When you select a single image, it now intelligently asks for your intent: you can choose a Detailed Visual Description to understand the scene or a Structured Text Extraction (OCR) for reading. The menu adapts dynamically based on the file type and your active AI engine.
• Core Optimization: We've performed a deep cleanup of the add-on’s internal logic, removing unused legacy functions and redundant code. This results in a leaner, faster, and more reliable experience for all users.

Hello Maurizio,
Thank you so much for your kind words and for sharing your detailed ideas with us!
I have great news: The exact feature you suggested has just been released in Vision Assistant Pro v5.5 (The Automation Update).
We have completely overhauled the "F" key (Smart File Action). Here is how you can use it for your specific needs:
1.
Activate the Vision Assistant command layer (NVDA + Shift + V).
2.
Press F and select your image file.
3.
The addon will now intelligently ask for your intent. Choose "Detailed Visual Description".
4.
Once the AI provides the initial description, a Chat Dialog will automatically open.
In this dialog, you can interact with the AI exactly as you described. You can ask follow-up questions such as: "I want to insert a 200x200 logo. Based on this image, what are the coordinates of the largest empty space where I can place it without overlapping any important elements?"
Since the AI analyzes the entire image and maintains the conversation history, it can guide you with high precision for your ImageMagick scripts.
Please make sure you have updated to the latest version (v5.5.2) for the best stability. I hope this feature helps you in your daily creative work.
Thank you again for your valuable feedback and suggestions!

Hi @JohnyTheHess,

Thank you for this excellent question! This is a very important use case for students and professionals in the accessibility field.

While Vision Assistant Pro does not yet have a dedicated "Scan" button for direct hardware control, you can achieve high-quality results right now using these two methods:

1. **For External/Document Cameras:** You can open the standard **Windows Camera app** (or any camera preview software). Once the document is visible in the live feed, simply press the **'O' key (OCR Full Screen)** in Vision Assistant. The AI will analyze the live screen content and extract the text immediately with high precision.

2. **For Scanners & Files:** If you scan a document and save it as a PDF or Image, simply press the **'F' key (Smart File Action)**. This will open a standard file selection dialog where you can browse to and select the document. The add-on will then perform an advanced AI OCR on the file directly, meaning you don't need to open it in a PDF reader or image viewer first.

As an instructor, your insight is very valuable. Would a simple "Capture from Camera" shortcut be sufficient for your students, or do they typically require advanced scanner controls (like DPI and brightness settings) within the add-on itself?

Thank you for your feedback and support!

Hi Johny,

It’s truly rewarding to hear that Vision Assistant Pro has made NVDA your top recommendation! That is the greatest compliment a developer in this field can receive.

Regarding the camera feature: We are definitely on the same page. My goal is to keep it "Point and Shoot"—no complicated DPI or brightness settings. Just a simple shortcut to grab the frame and present the text as quickly as possible.

To clarify the **'F' key (Smart File Action)**: When you press it, the add-on opens a file selection dialog. This allows you to browse and select any image or PDF directly from your computer. The AI then processes it without you ever having to open the file in a separate viewer like Adobe Reader or a photo app. It’s a great way to quickly "peek" into documents.

**Update:** I have just released **Version 5.6**, which introduces a new **"None (Extract Text Layer)"** engine. For searchable PDFs, this engine extracts text locally and almost instantly, which will be a huge time-saver for your students.

Thank you for your continued support and for being such a strong advocate for accessibility!

Best regards,
Mahmood

Hi Johny,

Thank you for the follow-up!

Regarding the documentation, the most detailed guide is built directly into the add-on. You can access it anytime by going to:
**NVDA Menu > Tools > Vision Assistant > Documentation**.

This will open the README file which explains all shortcuts and logic in detail.

To clarify the **"Automated Actions" (AI Operator - Shift+A)**:
Think of it as having a sighted assistant who can move the mouse for you. Instead of just hearing what is on the screen, you can give a direct command. For example:
* "Click on the 'Save' button."
* "Find the search bar and type 'Accessibility'."
* "Navigate through the installation wizard."

The AI analyzes the visual interface of the app, finds the exact coordinates of the element, and performs the click or typing for you. It is a game-changer for apps that have unlabeled buttons or are otherwise completely inaccessible to standard screen reading methods.

I hope this helps! Let me know if you have any other questions as you explore these features.

Best regards,
Mahmood

Hello everyone, the addon has been updated and you can get the latest version from the main post.

🌟 Support the Future of Vision Assistant Pro 🌟
Vision Assistant Pro is more than just an add-on; it is a mission to bridge the gap between AI and true accessibility. Every feature starts as an idea to solve real-world challenges for our community.

The Reality of Development:
Developing and testing a cloud-based AI tool in a region with extreme internet restrictions is a constant battle. To give you some perspective, a stable connection for development and rigorous testing often costs around $5 per gigabyte here.

Furthermore, cloud AI testing is not free. Running comprehensive test cycles for our visual features consumes active API credits—costing upwards of $10 per major test run out of my own pocket, which quickly accumulates over weeks of development.

Why Your Support is Needed:
To date, only three people have donated to this project. While I am deeply grateful for those contributions, more community involvement is essential to sustain the high costs of research, development, and testing infrastructure. Your support provides the resources needed to keep innovating and pushing the boundaries of what's possible.

How You Can Help:
Due to international sanctions, standard methods like PayPal are unavailable to me. I rely on the following alternative methods:

🍎 Apple US Gift Cards:
This is one of the most accessible ways to support. You can purchase them globally with low fees here:
🔗 https://www.mygiftcardsupply.com/shop/itunes-gift-cards/
📩 Please send the gift card code to: [email protected]

💎 Cryptocurrency (TON):
Wallet Address:
UQDoOOOoDYPP8eqWXVsjVyYzulY72JLZK1grPS_O2DbgVNsc

Thank you for being part of this journey and helping us make technology truly accessible for everyone!