Becoming Dauntless Newsletter: The Ultimate AI Interface

Written By Sofia Lazaro
The Ultimate AI Interface (and how you can capitalize on it)

But First…Cool, if Real

AI object recognition in real-time:

semantic vs instance segmentation

The is a real image segmentation tool by Ultralytics. Segmentation is Step 0 of what most people call object recognition. The segmentation tools are getting sophisticated. In the example above (read the whole post here) Ultralytics discusses semantic and instance segmentation. It's the difference between segmenting out horses in an image and segmenting out Horse A, Horse B, and Horse C. Both are applicable in different use cases.

Becoming Dauntless - AI Interfaces

The chat-style AI interface is giving 2002. There will be many ways of engaging with AI, but I think the one we will use every day will be in smart glasses.

For the Builders Out There, Here’s the Business Opportunity:

  • AI-powered productivity solutions

  • Real-time training tools for everything from pro sports to PowerPoint

  • Customer service or sales enablement overlays

  • Navigation, translation, note-taking, coaching, powered by voice + vision

  • New content experiences for influencers to provide followers

This is a blue ocean space. If you're a builder, designer, or entrepreneur, I don’t want you sitting this one out. The winners in this arena won't necessarily be AI engineers (although those skills will help) but those who know how to design for how humans experience AI. “Object recognition” has been on our product roadmap since day one (not hyperbole), and I’m beyond excited to start executing on that line item with Dauntless XR now that the AI tools are there.

Industry News

  • The Port of Corpus Christi is Getting a Digital Twin. Preach. We published our playbook on this shift back to digital twins (video podcast here), and you know I love a good dose of confirmation bias (almost as much as I love to see innovation in the maritime sector). I will add: digital twins make it easier to be hardware agnostic (2D devices, headsets, etc), enable multiuser modes beyond two people standing in front of a the same laptop. To maximize scalability and get digital twins out of the “expensive consulting” penalty box, you also need no-code twin editing and an agnostic AI model integration so you can switch up your AI as necessary.

  • Researchers Catalog 170+ Text Input Techniques to Improve Typing in XR. Only to find that multi-finger typing is the most effective method (duh). We need to create a brand new text input method. Voice-to-text is socially awkward, but that aversion is declining, and I think people will be A-OK with talking to their AI glasses. We’re glad someone is looking into it because the spatial typing experience is still terrible.

  • Google Shows Off Smartglasses with Gemini AI Continuing our convo on Privacy or XR: Pick One, the headset needs to record/track everything you look at to be the most useful. For example, AI smart glasses can tell you where you left your keys because it recorded and recognized what you were doing while you had the glasses on, including handling your keys. Like most web 2.0 examples, I suspect people will sacrifice privacy for convenience.

Saved on Socials

The posts, memes and videos that caught our attention and made us think this week.

Recently Played >

What our founders listened to this week.

Overheard on Slack

Enter the chat. What our devs are talking about…

  • If we were playing a drinking game with the words 'collaboration' or 'partnership' in this webinar, I think I'd already be dead, and we're only 20 minutes in. - When you get volun-told for a GovCon webinar.

  • Two wins in one day. Our emotional stock ticker is turning green. - March really knew how to deliver a punch to the face.

Like this newsletter? Subscribe via email to get it fresh and straight to your inbox.

Sofia Lazaro
Previous

Case Study: Technology to Modernize Aviation Maintenance
Next

Becoming Dauntless Newsletter: Privacy or XR - Pick One