Background

Since we’ve been incorporated in 2016, a lot has changed around us, whether that’s shifting hardware targets in XR, technology hype cycles (remember the metaverse?), or even changing acquisition patterns from within the DoD. But what has remained is our steadfast vision to quantify knowledge that moves missions forward.

We're excited to share with you today a new roadmap and vision for Holos that builds on our learnings from the past and turns it into a platform that's more capable, more agile, and more accessible than ever.

Our New Identity

You may have noticed an upgraded look and feel to Holos. Over the past few months, we iterated diligently on an identity that is more sophisticated, tied to our customer’s technical, maintenance-heavy roots, and yet future-forward. We wanted to create a design framework that is as exacting as the training that we help create, whether that’s for robots or humans.

Releasing our visionOS App

After developing on a new codebase internally for the last 12 months, we're incredibly excited to make the Holos app available on the Apple Vision Pro.

Holos on the Apple Vision Pro is an app that lets instructors create, update, and deliver technical training in spatial computing, without waiting months or spending millions.

We are especially proud of the fact that we believe that this is the first and only app that enables you create the training fully on the Vision Pro, without needing to use a separate device. We have focused intensely on making the Build experience as natural and intuitive as putting together a PowerPoint or Keynote presentation. Organizations can create their own courses, outline lessons, and populate them with any number of fully 3D stages. Stages can be built with a combination of pictures, video, text, and of course, 3D models. Check out the product page to learn more.

Since we've started in 2016, we have been watching (and oftentimes waiting) for the perfect device to come along to pair with our platform. We had always witnessed sizable friction from other devices, whether that was tethering to a PC, having to teach people how to use controllers, or having to limit the fidelity of our training content due to limited compute power. Naturally, when Apple announced the Vision Pro, we were already sold on the vision.

Apple is known worldwide for providing best-in-class hardware paired with incredible usability. The Apple Vision Pro allows us to push the limits in terms of fidelity, while also making it super simple to get people acquainted with spatial computing. We have always been a big proponent of hands-first, no-controller interactions, and having built-in hand and eye tracking has had a huge impact on how quickly our users can learn how to use Holos.

We're seeing this first-hand with out users at Luke Air Force Base. Adoption is quicker, and trainees feel more comfortable when they can see their environment in full fidelity alongside the virtual assets.

With Holos on the Apple Vision Pro, building and deploying immersive training has never been easier. If you have a Vision Pro, you can download the app today.

If you’re an organization that would like to pilot the full suite of functionality, please reach out here; we’d be happy to provide Trial access to credible organizations.

Holos Capture

Today, we're also announcing a new product in Private Preview for select Organizations.

Capture is an entirely new addition to our platform that lets users automatically capture performances of the physical processes that happen within their organization, directly on Apple Vision Pro. These organizations can then quantify and qualify human-led demonstrations into reusable datasets that can be transformed into skills useful for robotics and multimodal AI.

The origin of Capture can be traced back to a visit we had with our Air Force customers to Scale AI’s headquarters in San Francisco. Through numerous different discussions, we learned that the best, most credible insights during training come directly from the instructors that stand side-by-side with their trainees, not from the formal materials that sit on the shelf. In order to automate the creation of extremely complex tasks, we realized that we must formalize a way to automatically capture and quantify these insights.

After watching the humanoid robotics field make impressive strides over the last few months, we realized that the same problem exists for robots as it does humans, if not even more so. LLMs such as ChatGPT are able to operate with such precision because massive swaths of information exist on the open web. For robots who operate in the real, three-dimensional world, however, only a mere fraction of equivalent data exists. In fact, textual data used to train LLMs outstrips spatialized 3D data useful to robotics by a factor of 120,000 to 1.

Multimodal data is the foundation of embodied AI. The bottleneck to AI advancement is no longer compute or algorithms, but instead access to rich, high quality multimodal data that reflects how humans actually perceive and interact with the world. Without this grounding in physical reality, generalized AI models remain disembodied and limited to text and pixels.

Today, you can ask ChatGPT to tell you how to do something. But if you ask it to show you how to do something, it immediately falls flat. We see a future full of interactivity and demonstration, not just a voice that sits on your countertop or in your pocket. In order to progress the field, we believe it’s vitally important to have an automated way to capture rich 3D data for both fundamental and incredibly niche skillsets. That’s where Capture comes in.

With Capture, we're building the platform that captures, structures, and operationalizes this data, powering the next generation of embodied AI, AGI, and robotic systems with grounded, experiential knowledge.

It Takes a Village

We have numerous people to thank for making the launch of our new platform possible.

First and foremost, we want to recognize our incredible immediate team that hustled every day to make these changes happen. You should all be proud to have built something up from scratch, in a dev and design environment that had so many unknowns.

• Chris: Engineering Management, Unity Development, DevOps

• Alton: Unity Development, Swift, DevOps

• Surya: Unity Development, Swift

• Ankit: Unity Development, Swift

• Yaj: Swift, Authentication

• Rea: Holos API, Business Engineering, Authentication

• Jennifer: Identity, Marketing, UI/UX

• Saaketh: Capture Initiative

We can’t thank our partners across Air Education Training Command and Air Force Research Lab enough, especially MSgt Joseph Blahut at Luke AFB, for being our number one champions and for allowing us to be a part of such an important mission (read the case study).

We are eternally grateful to our board member, Mike Riegelman, our advisors, Jeff Moore, Mike Hurley, Ret. Major General Robert Wheeler, Carie Lemack, and Gene Cook, as well as our investors CEAS Investments, S3 Investment Partners, Paeonia Ventures, Techstars, Tundra Angels, and numerous Angels.

Special shoutout to Scale AI and Apple for being such supportive partners in our product and during launch.

We’re excited to iterate on our releases in public, and you can follow along here.