God’s Eye View: The Open-Source Spy Globe Bringing Palantir Power to Your Browser

The video opens with a provocation: what if anyone could run Palantir-like power on a home computer? The answer is Bilawal Sidhu’s God’s Eye View, a project that spent days at the top of GitHub trending, built viral momentum through a couple of striking demo videos, and now gets a source-level interview. Host Matt frames the intent with a Hollywood shorthand — those command-center dashboards that show where every satellite, plane and vessel is — and asks how to turn a static globe into a living place. The driving idea is simple: the world is already brimming with life, and open-source intelligence can bring that life to the browser if you stitch the right feeds onto a photorealistic globe.

What it is: a spy-satellite simulator in the browser

God’s Eye View is described as a spy satellite simulator in your browser where the underlying data is real. It borrows the visual grammar of a classified ops room — the same satellite, aircraft and vessel overlays you see in thrillers — and grounds every layer in public signals: flight transponders, ship beacons, orbital elements, seismographs and public cameras. The creator wanted to solve the interface bottleneck of OSINT: the signals are abundant, but the experience of them is fragmented across tabs. Turning those signals into a place makes them legible; the browser becomes a window onto infrastructure at scale. The positioning is explicit: situational awareness of large-scale physical systems, not tracking of individuals. You see where aircraft and vessels move, where satellites orbit, where quakes shake and where traffic jams — and you feel, as the host jokes, Google Maps for finding and Google Earth for getting lost, but now with an astronaut’s overview effect from your desk. Why play a fully synthetic game when you can explore the real world?

The audience is deliberately broad. Content creators use it to stage credible spy-scene shots with live data. City-watchers pull up their neighborhood and monitor local air and street layers. Journalists and activists reach for context without field risk; defense-adjacent viewers pop in because, in the founder’s words, people from that world say the civilian build looks cooler than theirs. The through-line is civilian intelligence: the tool now is not a defense product, but a public, legible alternative to opaque feeds. That is why the motto repeats — trade doom-scrolling for direct signal — and why the host keeps returning to the idea of democratizing what governments already use every day in conflicts. A single globe that aggregates what is already broadcasting invites collaboration rather than speculation.

How it was built: OpenClaw, Gemini and Opus in a fever dream

The origin story lands in March and April. The scaffold was raised by an orchestration harness called OpenClaw driving Gemini and Claude Opus together. The division of labor is presented as complementary: Gemini is strong at spatial reasoning, Opus is cracked at full-stack development. Even with a token-inefficient harness, the pair could lay the foundation and then layer satellites and aircraft one by one. The founder describes a fever-dream stretch of days where that was all he did, parallelizing work and stacking data sources onto the globe. The episode places that effort in a larger moment: a hoped-for bloom of a thousand flowers, now visible in the wave of flight sims and world-builders. God’s Eye View claims its own flower in that garden — a hackable foundation rather than a hardened production service, released under MIT with clear data-source and security notes.

Cost is engineered for accessibility. Outside the voice-interaction path that uses ChatGPT realtime and can get expensive, personal non-commercial use can run at no or near-zero cost; even all-day use is framed as a few dollars a month at most. Every layer exposes a get-key affordance in PowerUp, the provider-settings surface. You add TomTom for real traffic, AISStream for vessels, CelesTrak for orbits, USGS for quakes and so on — or you keep the free skeleton and explore the globe with what is public without keys. Anything needing a private key is brokered server-side, so the client stays local-first. No hidden scrape, no private dataset, no mystery pipeline — just public signal rendered where it belongs, on Earth.

Is it legal, can it spy on me: the hard line and privacy by design

The hardest line in the interview is privacy. The open-source build draws a boundary: no tracking of people. Period. What it offers instead is awareness of infrastructure: satellites overhead, aircraft in flight including some military traffic, vessels at sea, public cameras in select cities and aggregated traffic. The metaphor offered is a supercharged map that goes a couple of layers deeper than consumer apps. The counterargument is met head-on: surveillance tools already exist in governments and are in use in current wars; this project moves in the opposite direction by making the same public signals legible and inspectable. The goal is a window through which anyone can see the world that is already broadcasting, not a periscope into someone’s life.

Three design choices make that window trustworthy. First, delay: flights are rendered one polling interval behind realtime so the client can interpolate smoothly; public CCTV frames, as in Austin, update about one frame every ten minutes, so you cannot watch someone in realtime. Second, aggregation and anonymization: street flow is not a set of license plates but a per-segment measure of density and direction, color-coded like Google Maps where red means jammed, built from TomTom and OSM floating-car data. Third, labeling of simulation vs live: without a TomTom key, traffic is marked as simulated; camera poses are estimated until calibrated; launch ascents are tagged as reconstructed estimates. Every layer keeps source and freshness states visible — partial, delayed, simulated or unavailable — so users read the map with the model’s uncertainty in mind rather than mistaking a rendering for ground truth.

Live demo: from Atlanta to a San Francisco cockpit

The demo makes density tangible. Turning on flights, military flights, satellites and vessels at once fills the globe; the host stresses performativity — the client stays smooth while juggling thousands of moving tracks. A voice prompt — take me to the busiest airport in the United States — flies the camera to Atlanta Hartsfield-Jackson. From there the video cycles through presentation modes: 3D aircraft, landing approaches and a context mode that surfaces what else is nearby — neighboring planes, military tracks, vessels and mapped installations within a radius. Cockpit view then puts you on approach to San Francisco in realtime, a deliberately overkill way to follow a loved one’s flight if you wanted, but also a surprisingly pleasant ambient monitor while coding. That everyday use case — keeping a favorite airport live on a second screen — captures the project’s domestication of what once looked classified.

Military traffic earns its own beat in San Diego, a hub for major Navy and Marine bases. The palette tells the story: commercial traffic in white, military in amber yellow. Clicking a charter shows Dallas to Honolulu with clear operator and registration, while a nearby military track may read operator unknown and trace an odd holding pattern. The host asks how fresh the positions are; the answer is one step behind so paths interpolate without stutter. Vicinity awareness persists across layers: selecting a flight also lists nearby vessels and installations, each with stable identifiers. An aside demonstrates the agent’s world knowledge — DAL 346, registration N115DN, Delta Air Lines, plus a quip that Delta began in 1925 as a crop-dusting outfit — and then returns to collaborative exploration. Ask where the military activity comes from and the agent replies with training-base context, helicopter practice patterns and spacing drills. You could spend hours following those threads, which is exactly the point.

Down to the city: Austin traffic, CCTV and infrastructure

The camera then dives to city scale in Austin, Texas, orbiting the Capitol building with traffic and CCTV layers enabled. The coverage map reveals the mesh of publicly accessible cameras: each with an estimated pose and a slow-refresh snapshot. Traffic paints TomTom aggregates onto road segments; the encoding is familiar — red for dense, with flow direction — derived from floating-car data and anonymized before visualization. Google and TomTom both do this at scale with phones and connected cars; the project just surfaces the same aggregates on the globe. Beyond those two, the stack flexes: mapped installations, dams, submarine cables, data centers, and one especially revealing source — NASA FIRMS, the fire layer built from thermal satellites that flag heat anomalies above a threshold.

FIRMS turns heat into insight. Zoom near Ukraine and the border lights up with fire detections that align with areas of military activity, a grim but legible use of public thermal data available within hours of overpass, faster for the US and Canada. Earthquakes are plotted from the last 24 hours of USGS feed, with a sample event near magnitude 4.6 and a worldwide speckle of hotspots. Space missions round out the orbital story: a 30-day launch board lets you click a site, trigger a 3-2-1 countdown, watch an approximate ascent arc and settle into a rough orbit. The creator is explicit that the trajectories are not physics-grade — he lacks the compute and the degree to do exact propagation, and labels the playback as a reconstructed estimate — but for sense-making, approximation is enough. Each layer invites deeper dives rather than pretending to be authoritative.

Reconstructing reality: flood traces and skid marks

The deeper promise is reconstruction. The community has started using God’s Eye View as a base to rebuild the Iran flooding episode: satellite providers opened high-resolution imagery for the window, the flood’s course and steps were mapped on the globe, and user-generated clips from the ground were plotted along that same trajectory to play back the event in place. The host argues this is a better way to understand what happened than scrolling a feed of isolated clips on X; the before-and-after from commercial providers like Vantor makes devastation spatially legible. The tool’s role here is utilitarian — not a spectacle, but a shared substrate where satellite and street truth can be aligned. The same pattern extends to other hazards where open imagery is released for public interpretation, turning a scattered archive into a walkable chronology.

A second reconstruction is even more forensic: drone footage released by the NTSB, paired with ATC audio and ground-level imagery, is aligned to a 3D scan of a crash site to read the tarmac. The ATC exchange is audible, the aircraft’s rough path over the runway is replayed, and skid marks come into view precisely where the aerial and ground layers line up. You can hop between spots and keep aerial context while inspecting street-level detail. The host’s pitch sharpens here: between doom-scrolling and a single dot on a map with four people talking over it, can we offer a middle path — civilian intelligence that stays as close to ground truth as public data allows? That middle path, he suggests, is the point of the work and the reason to push it further, not as a surveillance fantasy but as a collaborative way to make sense of real events.

Install in two minutes: from GitHub to PowerUp keys

The closing section removes friction. The recommended path is to copy the GitHub URL, open the ChatGPT app or Claude Code, create a new local project called God’s Eye View, point it at an empty folder you control, and prompt the agent to clone the repo, read the instructions, install dependencies and tell you what remains to finalize. In the demo that loop takes under two minutes. The agent opens PowerUp and provider settings for you — also reachable as a plain browser URL — where each layer has a get-key link that jumps to the right signup. Out of the box you can already pan and zoom the globe, but the richer layers stay quiet until you add keys. The host finishes by flipping the switches in sequence: tens of thousands of commercial flights, amber military traffic, quakes, satellites and upcoming missions; then a voice request zooms to San Diego, toggles street traffic, swaps to CCTV, reveals data centers, draws submarine cables, lists vessels and highlights fire hotspots. The reflection that follows ties privacy and empowerment together: cameras and sensors are proliferating and personal privacy feels thinner, yet a tool that aggregates only public signals can restore a measure of agency. It feels like being a harmless spy with legitimate access — more understanding at your fingertips without stepping on anyone’s rights.