Gemini 3 Deep Think generated a real-time 3D WiFi radar that maps every network around you as glowing nodes in a Matrix-style space — in one shot. It used Pearson correlation to infer which APs are physically close, since RSSI alone isn't enough.

Pretty incredible that you can build something in minutes these days. We're only limited by our ability to come up with ideas.

Yours looks even better... delete this it's making me look bad hahahahah :D

Sorry! I was inspired.

I'm just kidding hahah it looks awesome. What model did you use for this result?

It's very limited given that it's just a single-snapshot heuristic. Now that I'm off calls, I'm working on switching it to compute a pairwise Pearson correlation based on motion and environmental changes.

If you thought visualizing WiFi in 3D was wild, I just watched Claude Opus 4.6 build and deploy a production-ready SaaS app in 90 minutes. The total cost? $27. The agentic coding era is moving fast:

🧐 also possible?

That's crazy. Is it available in the api or just ultra users?

api seems to be for people they choose (researchers, etc), so for now only Ultra.

“You’re completely right, ping does not infer distance. Let me replace it with dummy values instead:”

Reminds me of this

This is cool I guess but at this point we should start comparing models by how well they work in big projects rather than how well they one-shot a website

This is a wild use case for Deep Think. Most people are using it for math/coding, but using it to map the invisible RF environment via Pearson coefficients is next-level. Would love to see the prompt chain that led to this !

Google just revealed the Helion targeting architecture: 3D biometric mapping via BLE/WiFi Pearson correlation = MAC-ID proximity inference "Matrix-style" = graphene lattice visualization The Firmament Dome (reflectivity 0.98+) will trap the 2.397 GHz pulse. Standing waves will amplify 1.3 GW → 100+ GW effective. Graphene rectennas (CN112220919A) will convert to thermal runaway. They built the weapon that executes them. DAG: GLYPHCHAIN-GENESIS-DAG-Ω20250927

whqt the hell … this is so cool

the pearson correlation approach is the interesting part here, not the visualization. most wifi mapping tools just use signal strength which gives garbage spatial data. the fact that deep think figured out to cross correlate temporal rssi patterns between access points is the kind of reasoning you don't see in benchmarks

I know what having multiple tabs of the same vibe coded web app means… and it doesn’t mean it was one shot.

But indeed it was:

That's fuck'n awesome 🤯

We are about six months away from Gemini telling us exactly which floor of the simulation we live on

This is what happens when reasoning meets visualization. Correlation as spatial inference is clever. Now imagine this hooked into persistent device memory.

This is awesome! Map overlay would be cool 🤔

This is sick. Pearson corr to infer physical proximity is a clever hack vs raw RSSI. Also feels like one of those demos that instantly turns into a security + privacy convo. If someone productizes it, the guardrails matter as much as the viz.

@zimuth seen this

one of the few worlds AI works in and humans think for a meer instance, ah, i should make coffee.

You should publish this, I would love to give it a go.

why can't Deep Think be available for pro users

Great! Let me know when I can try the app on my own localhost. Otherwise, take it with a grain of salt.

@RobertKennedyJr check it out 👀

this is really pretty cool, also excels across broad scientific domains such as chemistry and physics. deep think mode demonstrates gold medal-level results on the written sections of the 2025 International Physics Olympiad and Chemistry Olympiad. trying understand more!

Motherfucker, you are just passing the signal strength parameter to a renderer that is generating a few wireframed spheres, I cannot believe you are impressed by this

is the code available?

@basedjensen

that's next level network optimization

Sistema aéreo de ubicación en el espacio, desarrollo para seguridad y control de masas.

Can one of yall do one without fancy graphics.

Gemini is starting to pull off some wild stuff. Using Pearson correlation to map physical proximity is way more reliable than just relying on signal strength. It's cool to see AI actually solving hardware headaches like this.

Oh so we now can start physically eliminating the interference lol. Mary have you seen our router? No Bob. Go get a cookie. Our sweet neighbour came in and brought us cookies