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

The author sorts which Dots system card numbers were measured, assumed, or never published, noting that an external red team got 1 in 12 attacks through while internal tests held 99.79%.

The author had Astra generate a Gaussian splat model of ChatGPT's frog-shaped Dots agent and added a WebGPU animation in the SuperSplat editor.

The author uses Grok Bot with existing content, competitor gaps, and decaying pages to decide what to draft, update, delete, or consolidate.

A user blocked by the Grok Bot usage cap complains there is no paid upgrade option, that SuperGrok or Heavy may not raise the Bot limit, and that the account link cannot be moved.

Someone released open source software for the Muse personal agent, claiming it lets you vibe code any smart gadget or new hardware product you can imagine.

A user comparing Grok, Muse, Manus Cue and other bots says they cannot use dots from Europe and asks to be added to the tester group.

Meta researchers had mathematicians use the Muse Spark model through the free public meta.ai chat to work on open problems, producing six papers that mark which passages were drafted by humans versus AI, including one case where a cross-field connection helped complete a proof.

A post claims several AI agents work together as one algorithm that repeats the same EMA-for-direction, RSI-for-momentum strategy 34 times an hour in BTC up-or-down markets, showing $692,556 in paper profit over 154 days.

In a reply, the user catalogs a long set of features they say are being missed, including Dots, workspace agents, GPTs, plugins, connectors, and deep research.

The author argues a personal assistant should be a single Dot, saying multiple Dots would confuse ordinary users with too many agents, modes, and models, and asks what people would actually do with many Dots.

Shub Gaur shows Grok Bots spawning sub-bots: one runs its own GrokBot account to create and test new bots, one replaces a 3–4 hour founder research workflow, and one keeps learning his writing voice.

The author argues dots is more than a personal assistant: spamming it with messages lets it spin up and manage multiple Codex sessions in parallel.