LIMA, now inside NVDA.
Hear what's on your screen, click a button just by describing it, and follow web pages as they change, each with a single NVDA shortcut. Free to install.
- NVDA+Alt+DDescribe screen
- NVDA+Alt+EDescribe focus
- NVDA+Alt+CClick by describing
- NVDA+Alt+WWeb narration
- NVDA+Alt+UCheck credit
AI for the screen reader you already know.
LIMA AI is a free NVDA add-on. Photos with no alt text, buttons with no label, pages that keep updating: it turns them into clear descriptions that NVDA reads aloud, so you're never left guessing what's there, all without leaving NVDA.
Describe your screen on demand
Press NVDA+Alt+D for a brief, factual description of what's on screen, handy for images, dialogs, and anything that isn't spoken.
Describe what you're focused on
Press NVDA+Alt+E to hear a quick description of the exact control or element you're on, without describing the whole screen.
Click by describing it
Press NVDA+Alt+C and say what you want, like "click the Send button" or "open the Downloads folder." LIMA finds that control on screen and clicks it, so you don't have to hunt for it.
Live web narration
Press NVDA+Alt+W to have web page changes read out as they happen, without interrupting your reading or repeating itself.
Know where your credit stands
Press NVDA+Alt+U to hear how much of your monthly LIMA credit you've used and when it resets.
Speaks your language
Get replies in English, Tagalog, or Vietnamese.
Built to fit into NVDA
No new app to learn. Your shortcuts are announced the first time you run it, and you can change any of them in NVDA's Input Gestures.
How it works
Three steps to get started
Install
In NVDA, open Tools then Add-on Store, search "LIMA AI", and install.
Sign in
Open NVDA Preferences, Settings, LIMA AI, and sign in with Google.
Use it
Press a shortcut, describe your screen, describe what you're focused on, click something by describing it, or turn on web narration.
Private by design
No API keys
You sign in with Google, and LIMA handles the AI for you.
Zero Data Retention
AI providers do not store or train on your screenshots.
On demand only
A screenshot is sent only when you press a shortcut, never in the background.
When you click by describing, LIMA uses on-screen accessibility information to find the right control, so a click only ever lands on a real control you can already reach.
Frequently asked questions
Is LIMA AI free?
Yes. The add-on is free to install and use, with a usage limit on its AI features. If you need more, the Go and Standard plans give you a larger allowance. Compare plans in the Roscommon shop.
What do I need to run it?
Works with NVDA 2023.1 or later on Windows. A free Google account is used to sign in.
Does it train on my data?
No. LIMA does not train on your data, and Roscommon Systems follows a Zero Data Retention policy for all of our products.
Will it interfere with NVDA?
No. The add-on is built to fit into NVDA. The shortcuts are announced the first time you run it, and you can change any of them in NVDA's Input Gestures.
What languages does it support?
Replies can be in English, Tagalog, or Vietnamese. You can change this in NVDA Preferences, Settings, LIMA AI.
Is it open source?
Yes. LIMA AI is open source under the GPL v2 licence. If you have used the add-on, you can leave a community review on GitHub.
Want the full experience?
The add-on brings LIMA's AI into NVDA for quick, on-demand help. For continuous narration and hands-free computer control, meet the full LIMA desktop app.
Explore LIMA