@kaseyklimes They are awful, but could be greatly improved. Currently revisiting Ted Nelson’s zigzag concept, but with a simpler, more linear interface.
@cpaik Token costs will continue to fall. Inference cost can come down with better workflow and context management. I don't vibe code so I'm not sure about what they charge power users, but don't you pay for tokens consumed in excess of the subscription fee?
@ankitkr0@cpaik Hmm... that's an interesting take. I agree that Chrome is advantaged here. Still room for new entrants to carve out their own niches. I would expect Comet to outperform Dia due to higher product velocity. I don't see perplexity slowing down at all.
@cpaik 1. agents can facilitate discovery and in doing so support leisure activities
2. opportunities to bring down inference costs abound; users are willing to pay for software they previously wouldn't have paid for
@ImadTaieb@karinanguyen Yes they are stuck in the old paradigm, but the browser does lend itself to new interaction paradigms. Both browsers are just copying chrome and cursor
@chrisahinojosa@joshm@diabrowser Yeah, it’s not as intuitive or frictionless as it could be. The browser should suggest the creation of skills when appropriate and they should be generated for the user. When the user is in a context where the skill would be relevant the browser should suggest it’s use