As we should be! thanks for commenting

In reply to Sarah Hoffman.

Exactly. Context matters more than any feature list. Glad the cognitive load angle resonated with you. Thanks for stopping by and leaving a comment.

Thanks for sharing your real-world testing experience, Michael. The tension between “doing” and “understanding” isn’t binary, it’s contextual. We’re exploring this further in upcoming pieces in our AI Browsers series, including one specifically on the security and reliability challenges facing agentic features. Stay tuned and thanks for stopping by.

I’m cautiously optimistic about both but skeptical of the automation promises. Great read, though!

The cognitive load section was great. As a UX researcher, I think we’ll need BOTH types of browsers depending on context. When I’m doing competitive analysis, Comet’s synthesis is invaluable and I haven’t caught it missing critical nuances yet though I always cross-check sources. But for repetitive admin work I’d love Atlas-style automation.

Great analysis on the trust issue. I’ve been testing Atlas for two weeks now and honestly, I don’t trust it to complete purchases yet. For now, I’m using it for research and letting it draft emails, but anything involving money? I’m doing that manually.