# Introducing re1: spatial memory for physical AI · Good AI Labs

> Search recorded events, check the source video and revisit captured places in 3D.

URL: https://www.goodailabs.com/blog/re1-video-memory/
Date: 2026-10-03

re1 is a spatial memory system for physical AI. Search recorded observations,
ask questions that lead back to source video, and inspect reconstructed places.
The recordings here show those paths in the current research preview.

## Find the observation

Search a recording. Check the answer. Revisit the place.

Captured in the re1 research preview.

1. Find the observation you need.
A search for machines handling packages indoors returns warehouse and robot-arm recordings. Open a match to inspect the source.
Figure: Actual search results. Open the image to inspect the original interface.
Capture: https://www.goodailabs.com/media/re1/search-results.webp

2. Follow an answer back to its source.
The captured answer cites 0:12–0:20 of the warehouse recording. The reader can open that interval and check the description against the video.
Figure: Excerpt from a recorded response. Its citation covers the stated interval.

3. Keep a place available after capture.
Move through a captured environment from a new viewpoint and inspect its objects. Prepared scenes remain available for later work.
Figure: A frame from the completed reconstruction shown in the video on this page.
Capture: https://www.goodailabs.com/media/re1/reconstruction-poster.webp

Start with a description of what you want to find. In this example, “machines
handling packages indoors” brings a warehouse and a robot-arm clip to the front
of the library. Open a match to inspect what happened.



The search view distinguishes confirmed matches from other candidates.
The image below is an actual capture from the product.

A query becomes a set of inspectable clips. For “machines handling packages indoors,” the captured search returns the warehouse and robot-arm clips. Unconfirmed candidates remain visible below the matches.
This capture shows frame verification at the search stage; the question workflow checks source video intervals.
Recording or figure: https://www.goodailabs.com/media/re1/search-results.webp

## Inspect what the answer used

A response is easier to check when it names the interval behind it. Here, a
request for a shift-log entry returns a description with a 0:12–0:20 citation.
The reader can return to those eight seconds instead of accepting an account
of the whole recording.

Follow the answer back to the recording. A captured response links its description to 0:12–0:20 of the warehouse video. It describes that interval; it does not establish safety across the recording.

An event absent from the cited interval may still appear elsewhere in the recording.
The source remains available for the reader to inspect.

## Keep the place available

The scene at the top of this page is a recorded view of a captured place in 3D.
The camera moves through that place, exposing
both its detail and incomplete regions. That scene persists after the
original capture.

Researchers can inspect objects and prepare scene artifacts for simulation.
Physical fidelity, including scale and material behavior, still needs validation
for the intended experiment.

## A memory that carries forward

Return to an earlier observation or reuse a prepared environment as the next
experiment takes shape. The [product page](/products/re1/) describes the current
capabilities and includes the recorded preview.

re1 is in development, with technical previews available. We are working toward
memory that helps researchers revisit experience, build useful simulation
artifacts and improve the data available for policy training.

Contact: research@goodailabs.com