Librarian or Butler?

An explorable guide to choosing AI agent types

There are two fundamentally different things you can ask an AI to do: find and present information (Librarian), or take action on your behalf (Butler).

The right choice depends on two factors: how costly errors are and how easily you can verify the output.

Drag the dot below to position your use case. The recommendation updates in real-time.


LIBRARIAN BUTLER DANGER ZONE HYBRID OK
Low error cost
High error cost →
Hard to verify →
Easy to verify

🔄

Hybrid Approach

Both modes may apply

Your use case sits in the middle. You might use AI for research and drafting, with human review before any actions are taken.

Recommendation


Why position matters

📚

Librarian

Retrieves, synthesizes, presents. Says "here's what I found." You decide what to do with it.

🎩

Butler

Executes and reports. Says "I've done the thing." You verify (or trust) afterward.

The difference isn't about AI capability—both can be sophisticated. It's about accountability and reversibility.

When errors are costly or hard to detect, you want the AI to inform your decisions, not make them. When errors are cheap and obvious, automation saves time without adding risk.


Your current position

Error Cost
Medium
Verification Difficulty
Medium