"AI Killed My Startup"

Cautionary Tales from the Tech Trenches

Author

Lori-Lee Elliott
April 27, 2025

But First…Cool, if Real

Google Glass in 2025:

This is real! Google stealth-demo’ed a pair of smart glasses running AndroidXR and Gemini at a TED Talk last week. The glasses look like normal eyewear and mercifully can have prescription lenses. Watch the TED talk here (YouTube). The forward-facing camera captures everything in your field of view for you to reference later, leading to some interesting privacy concerns I’ll save for next time.

Becoming Dauntless

"AI killed my startup" is trending across social media platforms. Google, OpenAI, Anthropic, and X routinely release free feature updates that replicate paid services built by aspiring founders, sending the fledgling apps to an early grave. How can you safeguard your venture and navigate our tech future with confidence?

Anticipate user behavior and build beyond the obvious next step.

Consider Kushank Aggarwal's experience with Map-This.com, declared dead (his words) after NotebookLM added a free mind mapping feature. In contrast, Poppy AI also offers mind mapping but with interchangeable AI models, plus functionality to create video scripts and content from those maps. They identified what a niche users set that would need several steps ahead, creating a value proposition that a single feature update couldn't easily eliminate.

At Dauntless, we're applying this approach. We recognize that people want seamless access to their preferred AI throughout the day, not just when they're on their phone. In parallel, smart glasses users demand context-aware information displayed through unobtrusive lenses.

This convergence of AI and XR is the future we're building toward. Smart glasses provide the ideal interface between humans and AI (at least until neural interfaces become mainstream), with built-in sensors that supply AIs with contextual information for truly personalized experiences powered by spatial computing.

This evolution creates opportunities for creatives and entrepreneurs to reshape our daily experiences.

Any predictive strategy involves making assumptions and accepting risk. That's where "fail fast" trope/methodology proves valuable. Make your assumption, take risks, but establish clear benchmarks to validate your hypotheses about user behavior. Meet your benchmarks? Proceed with confidence. Miss them? Return to ideation and pivot.

Industry News

  • Accelerating Industrial XR Adoption (Innovate Energy Now). A solid playbook on driving B2B adoption for emerging tech: get ALL stakeholder buy-in, select the right hardware, avoid programs with vendor lock, get the details and pitch nailed, and then put it in front of executives. Bonus tips: define success metrics up front, not just use cases, and track your metrics throughout (hint: metrics that matter are a combination of time/people/money). Also, check acronyms at the door if your grandma wouldn’t know what they stand for. Ex. POC can be proof-of-concept or point-of-contact.

  • 3D Models for Remote Maintenance (Breaking Defense). Gecko Robotics partnered with L3Harris for a remote inspection platform using volumetric scans that create a 3D model. The Navy is trying this with shipbuilding as well. The Dauntless team took this setup further with our apps Katana and Aura. A remote maintainer can write up a procedure on how to address an issue identified by these robots and send it to the maintainer onsite - no coding required. It’s the ideal interface for the deskless, hands-on workforce.

  • 100-inch Monitors in Augmented Reality. (MIXED). Another hardware startup came to the dark side and made the pivot to software (we have cookies, so I’m not surprised). If you have XReals and try this out, let me know.

Saved on Socials

The posts, memes, and videos that caught my attention this week:

Recently Played >

What our founders listened to this week.

Overheard on Slack

Enter the chat. What our devs are talking about…

  • "yay internet funnels work" - just when you start thinking that dead internet theory is true.

  • “Damnit, Mike, I’m supposed to be concentrating on YOLO-ing!” - when you’re building with an Ultralytics YOLO model for image segmentation and you keep getting distracted with new use cases. (YOLO = You Only Look Once, and it’s very cool tech if you haven’t seen it before).

  • "Someone thought a mechanical Turk was a profitability strategy instead of a prototyping strategy." - When someone takes a page from the Theranos playbook.

Like this newsletter? ⤵️