01 · Place
Pull your desktop into the room
Grab a real app window and anchor it where it belongs: eye-level, peripheral, desk-height, or across the room.
VentanaXR turns your computer into a room-sized spatial command center. Stream your real desktop into a headset, arrange apps as floating monitors, and use natural gestures to turn repeated work into AI-assisted workflows. The Quest 3 client now boots on real hardware, pairing takes four characters, and the live stream path is validated end to end.
Finance teams, developers, analysts, designers, and engineers already work across many surfaces: spreadsheets, terminals, dashboards, PDFs, CAD, inboxes, notes, and AI assistants. VentanaXR keeps that multi-screen muscle memory, but moves the workspace into mixed reality.
Your spreadsheet can stay life-size. Your terminal can sit below your model. Notes can float beside the document they belong to. The room becomes the interface.
VentanaXR should feel obvious in the first minute: windows become physical things you can arrange around the task.
Grab a real app window and anchor it where it belongs: eye-level, peripheral, desk-height, or across the room.
Curve dashboards into a cockpit, enlarge a drawing board, shrink chat into a side tile, or pin reference material next to the source.
Squeeze related windows together to summarize, compare, reconcile, draft, file, route, or create a repeatable recipe.
VentanaXR is built in the open with evidence. Here is what landed most recently.
The VentanaXR client installed and ran on an actual Quest 3 for the first time: 65+ minutes continuous, 72/73 FPS, 0.54–0.61 ms app frame time, zero stale frames or tears.
Production host configuration proved secure pairing, WGC capture, NVENC H.264 encode, and VTXP UDP delivery: 253 static datagrams plus 210 dynamic per-surface datagrams in 10 s.
The corrected Windows installer carries the real payload and waits for owner publish; the live v0.1/v0.2 releases still point at a placeholder build. Quest signing identity is the next release gate.
Direct D3D11 NVENC passed H.264, HEVC, and AV1 8-bit 4:2:0 plus H.264/HEVC 4:4:4 runs, with canonical validation and cyclic intra-refresh evidence.
The live capture-to-NVENC path now handles DPI-scaled and odd-dimension windows, validated on a real 2575x1407 window: 572 VTXP datagrams, 298 frames, 23 keyframes.
Use cases are shown as real spatial layouts, not generic feature tiles.
Model, GL export, variance dashboard, email thread, and notes stay visible while VentanaXR helps draft follow-ups and flag mismatches.
CAD, simulation output, issue tracker, terminal, and markup notes live around the object being reviewed.
IDE, terminal, browser preview, docs, tickets, and AI assistant can be arranged into a persistent build cockpit.
Papers, charts, notebook, browser, and citations become a wall of context that does not disappear when one tab changes.
Squeezing two windows together is not a gimmick. It is a shortcut for intent: compare these, draft from this, file this there, summarize these together, turn this layout into a recipe.
Spreadsheet + email → drafted explanation, mail merge, or exception list.
Window + destination → routed file with reversible metadata trail.
Repeated pairs become suggested automations you can accept or reject.
The MVP is a reliable spatial multi-monitor workspace. AI gestures layer on top once capture, decode, input, and layout feel solid.
Design around explicit window capture instead of indiscriminate room or camera upload.
Routing, filing, and drafting should show a reviewable trail with one-tap reversal.
Complex integrations can use cloud models, but the product should make that boundary visible.
What the codebase proves today, with gates that still need real hardware and provider acceptance.
No. The goal is to create virtual monitors from the host machine so the headset can display a large workspace without a physical monitor wall.
Quest 3 and Quest 3S. The client already installed and ran on a real Quest 3 at 72/73 FPS with zero stale frames for 65+ minutes; live streaming acceptance is the next milestone.
No. The strongest use cases are any work that depends on visible context: finance close, engineering review, research, analytics, design review, and software development.
AI is framed as a reviewed assistant: draft, compare, summarize, route, or suggest a repeatable recipe. The product copy emphasizes consent, visible boundaries, and undo.
The MVP is mostly built: the host stream path is validated, secure pairing is the default, and the Quest client boots on-device. Next comes in-headset acceptance of live pixels, input round-trips, and thermal behavior.
Yes. The visual sections are built so you can replace CSS illustrations with headset captures, videos, Figma renders, or product screenshots when available.
Yes. On August 4, 2026 the client installed and launched on a Quest 3 and stayed stable at 72/73 FPS with zero stale frames for more than an hour. Live streamed desktop pixels and input round-trips are the remaining acceptance gate.
Join the build list and tell us what kind of workspace you want to replace first.
No spam. Useful progress updates only.