Launching today
AI Pointer by Google DeepMind
Point at anything on your screen and talk to Gemini about it
2 followers
Point at anything on your screen and talk to Gemini about it
2 followers
AI Pointer lets you point at any element on your screen and ask Gemini about it using speech or shorthand, without switching apps. For anyone who works across browsers, docs, and tools and wants in-context AI help.
The mouse pointer hasn't meaningfully changed since 1968. Google DeepMind is running experiments on what happens when it does.
What it is: An AI-enabled pointer that reads the context around your cursor and lets you interact with Gemini using natural speech and gestures, without leaving the app you're working in.
Current AI tools sit in their own windows. You have to pull content into them: copy-paste a paragraph, describe an image, explain which table you mean. The AI Pointer flips this. It sees what's under your cursor and resolves references like "this" and "that" without requiring a detailed prompt.
What makes it different: The combination of cursor position, visual context, and voice works together to resolve intent. You don't write a prompt describing what you want to act on. You point at it and speak in shorthand. The system infers the object from the pointer position, not from your description of it.
Key features:
Cursor-aware Gemini integration that captures visual and semantic context around the pointer
Voice plus gesture input: say "fix this" or "double these" while pointing at the target
Works across apps in the browser without switching context
Pixels-to-entity resolution: a photo of a note becomes an interactive to-do list, a video frame becomes a booking link
Live experimental demos tryable in Google AI Studio today
Benefits:
Removes the need to write precise prompts for on-screen content
Keeps you in the workflow you're already in
Reduces the copy-paste-describe loop that current AI tools require
Who it's for: Knowledge workers, developers, and technically curious users who already use Gemini or Google AI Studio and want to explore what pointer-native AI interaction looks like in practice.
The deeper bet here is that the text prompt box is not the final interface for AI. What DeepMind is sketching out is a layer where intent is inferred from context rather than typed. Whether this ships as a real product or stays as a research direction, it's a useful thing to have in front of the PH audience.
I hunt the latest and greatest launches in tech, SaaS and AI, follow to be notified.