Kill list

These fail the loop test. They are content mills, marketplace spam, or undifferentiated services. Do not spend a test week on them.

A bot-native business needs a bounded event loop: watch → enrich → decide → act → escalate. The items below optimize for volume of generated artifacts or vague “we automate things” sales. That is not the same job.

Faceless YouTube farms

Kill. The hard part is distribution and platform policy, not generating another video. There is no customer-specific watch surface, no decision rule a buyer pays for, and no escalate path that turns into an ops contract. Treat “faceless channel at scale” as a platform-risk hobby, not a CRO bet.

SEO spam

Kill. Publishing undifferentiated pages for search crawlers is not a watch-enrich-decide loop for a paying operator. Quality systems and manual actions are the counterparty. If the “customer” is a search index, you do not have a buyer.

KDP floods

Kill. Mass-uploading generated books into a retail catalog is the same shape as SEO spam: flood a marketplace, race to the bottom, absorb TOS risk. No bounded event, no filtering product, no escalate-to-human ops motion worth selling.

Generic AI content agency

Kill. “We will write your blogs / ads / social with AI” is a services company with a cheaper intern. Sales, taste, and account management dominate. There is no single event stream and no productized decision rule. If a content motion appears later, it should be inside one of the surviving ops loops (for example, review replies), not a standalone agency.

Vague automation agency

Kill. “We will automate your business” has no watch surface and no default action. Every engagement is a custom project. Bots can exist inside a named loop (RFP scout, review ops, doc-to-data). They cannot exist as a generic automation promise.

Not on the kill list — just weaker

Wholesale real estate is not in this kill bucket. A bot can help with public-data research and memos. Bot leverage is still ~2/5 because phone and trust close the deal. Rank it below the top five ; do not start there if the goal is a bot-native loop.

Sources

ChatGPT Deep Research export — TBD from Google Drive Research/Inbox.
CRO-approved kill list (no extra “dead” ideas added).