What we won't automate with Grok Bot (and why)

What we won't automate

Trading, money movement, posting as a person, outreach at volume, login-challenge workarounds, and AI interviews. The list, with reasons.

Short answerIf it trades, pays, posts as a named person without approval, blasts strangers, defeats a login challenge, or interviews humans, we don't build it. Not because it can't be built; because the failure modes land on you.

Trading and money movement

An always-on agent with a brokerage login, no spend cap, and exposure to prompt injection from every page it reads. We won't. A read-only research Bot that summarises filings and flags alerts for a human is fine; execution is not.

Posting as a named person

Company accounts via API with approval, yes. A Bot on your personal LinkedIn or X session, no. It's your reputation and your account restriction.

Outreach at volume

Forty suppliers or four hundred prospects contacted by a Bot is the thing recipients are already building filters against. It burns your domain and your name. Follow-ups on existing relationships, drafted for approval: yes. Cold blasts: no.

Login-challenge workarounds

The Bot's browser will be challenged. The answer is API access or a human step, not techniques that break next week and violate the site's terms.

AI interviews and automated candidate screening decisions

Summaries for a human to read, yes. A Bot deciding who gets rejected, or an AI "interviewing" candidates, no.

Anything that needs your personal logins

If it truly needs your identity, it's your task.