AI Tells
AI writing has recognizable patterns that trained readers spot quickly. These are not errors exactly, but habits the model develops from training data that cluster together and signal machine authorship. Each control below describes one pattern, shows what it looks like in practice, and lets you dial it back or remove it from the output entirely.
Leaving a tell at No preference means the model decides; selecting Minimize or Eliminate adds an explicit instruction to the generated modifier block.
Generated Response Tuning
Paste this block at the end of any prompt.
If you find yourself copying the same modifier block into every prompt, or if you have business context that applies to most of your AI interactions (your role, your audience, your brand voice), consider storing it somewhere more permanent.
Most AI tools have a container for this: Claude and ChatGPT call it a Project, Gemini calls it a Notebook. Paste your tuning instructions into the system prompt or context field of a project once, and every chat inside that project inherits it automatically.
This works especially well if you have a consistent use case like drafting marketing copy, writing internal documentation, or responding to customer emails where tone, audience, and register rarely change.