Use · ai workflows
AI workflows: a human review gate at every irreversible step
How Less Typical uses AI as a comparison surface rather than an answer machine, with explicit human gates at the points where words become action.
AI workflows
Less Typical uses AI. It refuses to be used by AI.
The rule
AI increases output. Less Typical increases judgment.
This is not a slogan. It is a workflow constraint.
Any step where AI output becomes an irreversible real-world action — publishing a post, sending a sales email, marking a student paper, recommending a hire — must pass through a human review gate.
The pattern
1. Frame the request as a default + a structural alternative + a critique.
2. Generate candidates.
3. Run each candidate against a real reader, student, buyer, or case.
4. The human chooses the one that survives.
5. Publish only after a real reaction, not a self-flattering rewrite.
This is the operational form of the AI as comparison surface move and the minister, not sovereign Leitwort.
Anti-patterns to refuse
- Auto-publishing: Letting AI draft and post without a human gate. The discipline dies here.
- Polished-as-progress: Rewriting until the AI output “sounds good” without testing it.
- Single candidate trust: Asking AI for “the best answer” and taking it. The tail-space move is structural difference, not louder confidence.
- Mimicry: Asking AI to write “in Brent’s voice” without specifying the actual handle behind the voice.
When to use AI anyway
When the move is hard:
- generating many candidates before selecting (verbalised sampling);
- probing the shared assumption behind a polished answer;
- rehearsing an objection before a real conversation.
When AI is the wrong tool:
- when contact with a real reader is the only thing that can test the handle;
- when the move is about presence, not production;
- when the discipline requires slowing down, not generating more.
Related moves
ai as comparison surfacedifferent correct useful