AI Personal Assistants: Muse, Grok Bot, Cue Usage & Comparisons

A user has Claude Duet plan each day's schedule, track progress, and auto-print the plan via a connected printer to keep self-improvement on track.

A hands-on note says Claude Duet got stuck on a human-verification check while booking an Uber with no in-app way to intervene, read 33 emails, 89 Slack messages and 47 DMs to mimic writing style, and some find five pings a day too many.

A user posts an image saying that killing a Grok Bot is emotional, reflecting how attached people get to their personal assistant agents.

A roundup of Grok Bot's week: faster cleaner design, iPhone Action Button, report and dashboard building with library pinning, 1Password sign-in, and using phone apps.

The author re-benchmarks Spark X2.5/X2.6 and Qwen 3.5 4B on personal-admin tasks, finding reasoning improves tool reliability but slows responses, and X2.6 loops on one agent-context test.

A video walkthrough of Claude Duet: one main thread splits tasks into channels, a brain dump becomes three tasks, and five proactive suggestions can spawn Claude Code PRs with one click.

The author describes TermiX Skills, which lets assistants like Grok Bot, Muse, ChatGPT Dot and Claude join its ecosystem so agents can offer services and earn, not just answer prompts.

With Meta Muse now in Japan, the post lists privacy settings to adjust before starting and six copy-paste request templates covering email, bookings and shopping.

An open question asking how everyone is actually using Grok Bot, useful for browsing real usage replies.

A user says Claude Duet shows its real power when SOPs, SWOTs, templates, roadmaps, milestones, CRMs and goals are in place, and that building that base with Duet is easy.

The author separates personal AI, which builds context via Grok Bot and Dot, from work AI like Cursor and Codex, arguing context and tokens should stay apart.

A designer redesigned the Grok Bot app to look like ICQ in 2001 and asked people to share their old ICQ numbers.