Writing Laboratory
EN PL
Let's Connect

Methodology

The Writing Lab

Structured multi-perspective debate, AI as instrument, human judgment as director.

Version 1 · February 2026

What it is

The Writing Lab is a method for structured thinking. I assemble panels of distinct perspectives — each prompted from a specific thinker's published work — and orchestrate debates between them using AI. The goal is collision, not consensus: surface the assumptions each framework takes for granted, illuminate gaps no single viewpoint can see.

This is the same skill I use in executive workshops with 40+ managers: design for productive disagreement, protect the minority view, drive toward synthesis without forcing premature convergence. The AI makes it scalable. The structure makes it rigorous.

Where it comes from

The Writing Lab draws on Dr Paweł Szczęsny's research on psychological modulators at Neurofusion Lab — the insight that persona prompts don't just change an LLM's style but activate distinct reasoning pathways already present in the model. A single well-chosen persona unlocks capability that default behavior buries.

The Writing Lab extends this from single-persona steering to multi-persona dialogue: if one perspective unlocks one reasoning pathway, structured collision between perspectives illuminates what each one can't see alone. The value isn't in any single viewpoint. It's in the friction between them.

How it works

1
Frame the question

Every discussion starts with a question worth arguing about. Not "what should we do?" but "what tension exists between these competing perspectives?"

2
Assemble the panel

Select 3–6 perspectives that will create productive friction. Each is prompted from a specific thinker's published work — the specificity of the source produces the distinctiveness of the voice.

3
Orchestrate the debate

Structure the conversation in acts: diverge, collide, converge. A facilitator (also AI-generated) manages the flow. Participants enter and exit based on expertise. The director — me — shapes the whole.

4
Harvest the insight

What survives the collision is stronger than what any single perspective could produce. The recommendations have been stress-tested. The gaps have been illuminated.

The roles

Director
Conceives the question, selects the panel, shapes the arc, edits the result. The human in the loop. In these discussions, that's me.
AI instrument
Generates the perspectives based on directed prompts. Currently Claude (Anthropic). The tool, not the author.
Participants
Fictional personas — each informed by a real thinker's published work, presented through descriptive names. "The Complexity Scientist," not the person who inspired them.
Facilitator
An AI-generated role that manages conversation flow — names tensions, drives toward synthesis, prevents premature agreement.

The persona system

Real thinkers' names and published work are used during generation — because an AI prompted with a specific body of work produces something meaningfully different from one prompted to be "a complexity scientist." The specificity matters.

But in publication, names become descriptive personas. "The Product Strategist" instead of the person who inspired them. This protects real people from false attribution while preserving the intellectual shape of each perspective.

Every discussion credits its sources in a dedicated attribution section. The reader knows exactly where the ideas come from. They encounter a role, not a performance of a specific person.

One exception: literary figures whose names are inseparable from their imaginative universes — Lem, Borges, Pratchett — may keep their names when the function is creative rather than authority-borrowing.

Panel rules

  • Maximum 6 participants simultaneously in any discussion
  • Participants rotate based on expertise — arrivals and departures are noted in the text
  • Same thinker always appears under the same display name across all discussions
  • Every discussion is unmistakably identified as AI-staged before the reader engages
  • Every discussion points readers toward the original thinkers' work

Commitments

  • Transparency. You always know this is AI-generated, directed by a human. Before you read a single exchange.
  • Attribution. Every thinker whose work informs a perspective is credited, with pointers to their original writing.
  • Accountability. My name is on this work. If a thinker whose work has been invoked believes they've been misrepresented, I want to hear about it.
  • Invitation, not imposition. The persona system is designed so real thinkers can engage, respond, disagree — without the awkwardness of reacting to a puppet version of themselves.

See it in action

The Writing Lab is a practice, not a product. I use it in my Laboratory and in my advisory work. You can use it anywhere. The method is open — what matters is the commitment to transparency, attribution, and structured disagreement.

If you do use it, I'd like to hear about it. [email protected]

Created by Piotr Kuczyński · Innomada · Warsaw