yisrael_g

Hi HN, I built Moe, an assistant that lives on the notch of your Mac instead of in a browser tab, terminal, or app.

Most AI tools are a text box. With Moe you talk to it out loud, it does the 'thing', and it keeps what you've told it, so you don't repeat context every session.

What it can do: Connect to your personal or professional life (e.g. Gmail, WhatsApp, Notion etc.). It has it's own browser on your machine, your account, your logins, no interruptions. And it can do anything you need done that is within the digital limits of a Computer (AKA Everything).

Happy to answer anything about how it works.

Each engagement required feedback, the quiz is just a layer, the mandatory feedback is the real bottleneck to get around. 3 questions about the product, and many layers of security that requires a human to have knowledge of the product, we detect any LLMs attempt to enter the fold. So even if the user tries to cheat, and he will only achieve that up to a certain point. But what the company pays for is to put their product in front of someone, and get proof it was understood many X more than a click is "understood".

In other words by a user providing feedback he is proving to a degree that he "understood" the product. That is what he gets paid for, and that is what the companies pays for.

Identity verification, and biometric verification (during each engagement), are the main two. For the full list: https://www.nexertise.com/security

It's true, but the system seems to be running the way it is for decades. What's the proposal?