Prompts Are Conversations You’re Not Present For

Source: Limited Edition Jonathan Substack
Author: Limited Edition Jonathan
Original source: https://limitededitionjonathan.substack.com/p/prompts-are-conversations-youre-not
Source type: essay

Private backup: the full article text is archived in the private repository at archives/articles/limitededitionjonathan-substack-com-prompts-are-conversations-youre-not.source.md. It is not published on the public Quartz site.

Summary

The author argues that one underused form of AI work is designing prompts for other people to run, not just prompts for oneself. In this model, a prompt becomes a mediated thinking process that structures another person’s reflection, extracts context, and produces usable artifacts. The central example is an operating agreement for AI Cred: instead of negotiating from fixed positions, the author and partner each used a structured AI-guided question sequence to clarify their own reasoning before comparing outputs. The article offers three design principles for prompts-for-others: minimize cognitive load, use Q&A context extraction, and produce immediately usable artifacts.

Big ideas

Claims

Key evidence and examples

  • The AI Cred operating agreement example uses Claude to walk each partner through decisions one question at a time.
  • Outputs from two people running the same structured prompt were compared to identify alignment and negotiation points.
  • Design principles include minimizing cognitive load, extracting context through one-question-at-a-time dialogue, and producing usable artifacts.
  • The article frames a prompt as a thinking process designed for another human to run without the author present.

Pull quotes

Prompts as thinking processes

Here’s the thing that made this work: the prompt wasn’t instructions for AI. It was a thinking process I designed for another human to run without me in the room.

Questions create clarity

A well-designed prompt isn’t just collecting information. It’s guiding someone through their own thinking process. The questions create clarity that didn’t exist before.

Designing another person’s thinking

You’re not writing instructions for AI. You’re designing how another person will think through a problem - and making sure what comes out the other end is something they can actually use.

Education relevance

Relevant for AI literacy, project-based learning, teacher workflow design, student collaboration, writing pedagogy, peer feedback, research intake, advising, and administrative planning.

Durability note

The examples are tied to current AI workflow habits, but the central design lesson is durable: a good prompt can encode a reusable thinking process for another person, not just a one-off instruction to a model.

My notes