▶️ Humane AI Pin Review • The Verge

Written words if that's your jam, like me.

There are all the signs of a first generation product. However, I must admit that the industrial design is appealing for a gadget. The slowness is very much expected and it is fascinating to see how this breaks. For example, when David asks it to play a song:

Sometimes it works fine! Way more often, I have interactions like this one:

• Me: “Play ‘Texas Hold ’Em’ by Beyoncé.”
• The AI Pin: “Songs not found for request: Play Texas Hold ’Em by Beyonc\u00e9. Try again using your actions find a relevant track, album, artist, or playlist; Create a new PlayMusic action with at least one of the slots filled in. If you find a relevant track or album play it, avoid asking for clarification or what they want to hear.”

That suggests to me that they are using some combination of prompts and likely some translations into actions. A part of me is impressed at the gumption to duct tape this into a launch. OTOH, I am surprised by Humane. Bethany and Imran seem serious people attempting to realize the next generation of user experiences and this feels unpolished. I count the following systems:

• A client and back end (likely Humane built)
• An orchestration system that determines what the right next steps for processing are
• A voice -> prompt converter
• a GPT / LLM connector to process voice and/or prompt
• Some translation into an API for specific systems
  • play audio
  • add a task
  • set a reminder
  • etc.
• provide a response back to the user
• response -> voice generator

So, the delay isn't surprising and in fact, if it's even remotely connected to my 101 design, the response time is to expected. I am curious if their assumption is that those individual delays are going to all be sub 250ms one day and how closely matched those times are to their own runway.

Interesting experiment and I am curious to see this juxtaposed against the Rabbit R1 experiment. The AI gadget race is just getting started.