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

A long Chinese post breaks down the five tiers of The X Subscription and argues the $30 tier, which includes Grok Bot and AI Agent, is the most appealing.

A demo video of Grok Bot, captioned that it has entered the Matrix, showing the personal assistant operating a computer on its own.

A guide uses the JLC ESP32-S3 board with Muse to ask questions by voice and get text plus voice replies, with code and docs open-sourced on GitHub.

The author tells Grok Bot what they want, and it passes the work to a Cursor Cloud Agent that researches, writes code and opens a PR with a live preview, with nothing merged until the author approves.

A user says that after posting a WhatsApp status, Grok Bot wrote and published it to X, LinkedIn, Instagram and YouTube.

The author argues Grok Bot should only gather context rather than orchestrate, leaving scheduling to a Coordinator Agent in Projects, since understanding a problem needs less intelligence than on-site execution.

The post says Google Gemini Spark turns AI from a chatbot into an around-the-clock personal agent and lists 8 prompts to make an AI agent actually useful.

Meta Muse Wonder fully solved or disproved 44 of 226 research-level math problems (19%) on ErdosBench, ranking fifth.

The Qwen app launched a limited invite test of a personal assistant built on Qwen 3.8's agentic capabilities, with users reporting fully automated flows such as ticket booking and delay claims.

The author compared Meta's two personal assistant models with a peacock test, and the underdog Muse Spark 1.3 Max came out on top.

With Grok Bot now available on lower tiers and Claude Code handling more of his coding, the author moved from Cursor's $200 Ultra plan to the $60 Pro+ plan.

The team says suggestions triggered by a new email, a calendar change or a periodic check are free, but work carried out after a user confirms is still billed, and recommends connecting email, calendar and Slack.