Hark
At Hark, we are building the most advanced personal intelligence in the world.
ok wait, "verifiably the best" is a big claim to lead with. drop the eval suite or it didn't happen.
hot take: benchmarks are cool but I want to see it survive a Delta reschedule with a basic economy fare. that's the real eval.
how many people shipped this? genuinely curious because the demo scope feels like a 40 person team but the vibes feel like 6.
the cut at 0:14 where the agent silently opens three tabs did more marketing work than any voiceover could. whoever edited this understood the assignment.
would love to chat about the training setup for the browsing model whenever you have a minute. happy to keep anything off record if easier.
building something adjacent in the shopping agent space and honestly the handoff framing is smart. slightly annoyed I didn't call it that first.
one more thing for the roadmap: let me hand off a task mid-execution to a human and back. the name is literally begging for it.
3 weeks out from launching my own agent thing and every time I see a Hark drop I add another week to my timeline. this is not sustainable for my psyche.
the pacing on this launch is unreal, three category releases in a row and each one lands cleaner than the last. hark_labs cooking.