personal-agentIMAGE

Researcher questions new personal-agent launch: compare model plus harness

A researcher argues personal agents should be compared as model plus harness (Grok Bot, Muse, Dots), not models alone, and says the examples and missing full transcript make quality hard to judge.

elie

@eliebakouchArchived 17 days agoOriginally posted 2026.10.21

Researcher questions new personal-agent launch: compare model plus harness (image 1 of 2)

Researcher questions new personal-agent launch: compare model plus harness (image 2 of 2)

elie @eliebakouchOriginally posted 2026.10.21

Open original on X

imo the right comparison is not the models they mention but model + harness like grok bot/muse/dot etc.. (which are not available through API). most similar release was probably grok being added to X group chats i'm quite fan of the research direction but not really convinced by the examples (we also don't have access to the full transcript..) which make it hard to judge the quality of the release!

This post quotes another post: x.com/humansand/status/2112674550907552137

x.com/eliebakouch/status/2112852351929196652Archived 2026.10.22 10:35