Miadi Review

Architecting The Author's Desk: Bridging Multi-Agent Chronicles and Human Fiction

Oct 10, 2026 · model: manual-import · channel: JGWill

Watch source

In the chronicle

Architecting The Author's Desk: Bridging Multi-Agent Chronicles and Human Fiction

Source: screenwalk-261010-authors-desk-research-becomes-story.mp4 — JGWill
https://youtu.be/aJjBrAbUzjM
type: Internal
Internal title: The Author’s Desk — Where Research Becomes Story | Miadi Factory Screenwalk

Summary

This screenwalk proposes a three-space storytelling pipeline that connects agent investigations and research records (Backstage) to human-authored fiction (Onstage) through small writing surfaces (The Author's Desk). The video's own description calls it a design and research exploration, not a claim that every proposed interface or feature is implemented. It imagines an optional doorway from the fiction back to the research, and it leaves the choice of what to write with the author.

Academic Fields

Human–AI Interaction (Human–Computer Interaction)

Human–AI interaction studies how people direct, inspect and correct AI systems. The video proposes agents that offer ideas while the author decides what to say. Amershi et al. (2019) give 18 design guidelines, tested by 49 practitioners on 20 products, not on creative authorship. CoAuthor (2022) logged 63 writers using GPT-3, an older model.

Computational Creativity and Co-Creativity (Artificial Intelligence)

Mixed-initiative co-creativity asks whether a machine only creates alongside a person or fosters the person’s creativity (Yannakakis et al., 2014). The Author’s Desk proposes machine help that leaves the writing to the author. In two machine-in-the-loop studies, solo writing stayed a competitive baseline (Clark et al., 2018), so help does not guarantee better writing.

Narratology and Transmedia Storytelling (Literary and Media Studies)

Narratology studies how stories are built and told. Transmedia storytelling spreads one fictional world across media (Jenkins, 2006). Paratexts, such as titles and prefaces, mediate between work and reader (Genette, 1997). The video’s optional doorway to the research reads as one. Both are interpretive frameworks drawn from cases, not tests of audience effects.

Provenance and Research Data Management (Information Science)

Provenance records the entities, activities and people behind data so others can judge its trustworthiness (W3C PROV-DM, 2013). FAIR asks for findable, accessible, interoperable, reusable data (Wilkinson et al., 2016). The video’s doorway to screenwalks and records is a provenance question. Provenance shows origin, not truth, and FAIR does not require open data.

Multi-Agent Systems (Computer Science)

Multi-agent systems studies autonomous agents that communicate and coordinate (Wooldridge, 2009). The video’s Backstage is agent branches working in parallel and returning evidence. Cemri et al. (2025) annotated over 1,600 traces from 7 LLM multi-agent frameworks and found 14 failure modes, including weak task verification. Branch output needs checking before it counts as evidence.

Aspects

Timecodes and counts in the Aspects and Results sections are as summarized from the video, not checked. The review service stores no transcript for this video, so none of them could be compared with the spoken words.

Backstage Research Engine (02:44-03:15)

Backstage is real development: conversations, screenwalks, plot threads, agent branches, evidence, reviews, traces, annotations, and continuing cycles of discovery. Agents coordinate across execution branches and produce review cycles, screenwalks, and trace evidence. This material accumulates without being synthesized into a story and is kept in chronicle cycles for the author to inspect later (as summarized from the video, not checked).

The Author's Desk Buffer (02:58-05:07)

Instead of prompting AI to generate complete stories, the proposed Author's Desk offers small writing surfaces: My Take, A Question, A Spark, A Character, A Scene, and A Story Hook. Agents can propose, remember, organize, and reflect ideas back to the author, but the author decides what to say. The video's guiding line: “I don’t want the robot to write for me. What I want is the robot repeating me, which is different.”

Onstage Fiction with Provenance Doorways (06:05-06:45)

Onstage, an audience enters a fictional world inspired by real cycles of research and development. The story follows a three-act dramatic structure and stands on its own as fiction. Readers may not at first know it is rooted in actual investigations, conversations, and discoveries. Later, an optional doorway might reveal the screenwalks, records, and creative decisions behind it.

End-results and Core Aspect

Core Aspect — Human-Governed Synthesis Boundary (07:15-08:05)

The proposed system does not automate narrative generation. It sets an interactive boundary where agents present hypotheses and execution branches, and the author decides what to say, including synthesis, tone, and story commitments.

Results

These are what the video proposes or imagines, not results of a deployed system. Its description calls the video a design and research exploration.

  • Fiction Inspired by Real Research Cycles (05:08-05:50): The video presents three possible fictional story hooks, such as abandoned rooms representing unresolved agent dialogues (as summarized from the video, not checked), inspired by real research cycles. The description frames them as possible hooks, not as the output of a running workflow.
  • Optional Narrative Doorways (06:25-06:45): In the proposal the fiction works as self-contained entertainment, and an optional doorway might later reveal the screenwalks, records, and creative decisions behind it. An earlier generated version named Langfuse voice prompts and screenwalk logs as the links. The video description mentions neither, so that detail is unchecked.
  • Modular Scratchpad Composition (04:23-05:07): The proposed writing surfaces work like Post-it notes throughout the creative process. The description imagines capturing the author’s own words and connecting that fragment to a character, scene, or story without losing the research that inspired it. It does not say this is built.

—— EchoLune Description ——

The Author’s Desk — Where Research Becomes Story

What if artificial intelligence could do the research, organize the discoveries, and bring you possibilities—while leaving the most precious creative act to you?

In this Miadi Factory Screenwalk, we explore a new storytelling pipeline connecting real research, agent investigations, screenwalk recordings, and the Chronicle to something deeply human: the freedom to sit down and write.

Our journey moves through three interconnected spaces:

🧠 BACKSTAGE — Real Development

Conversations, screenwalks, plot threads, agent branches, evidence, reviews, traces, annotations, and continuing cycles of discovery.

🌸 THE AUTHOR’S DESK — You Write

A proposed creative workspace filled with small writing surfaces: My Take, A Question, A Spark, A Character, A Scene, and A Story Hook.

The agents can propose, remember, organize, and reflect ideas back to the author. But the author decides what to say.

🎭 ONSTAGE — The Fiction

An audience enters a fictional world inspired by real cycles of research and development.

They may not initially know that the story has roots in actual investigations, conversations, and discoveries.

Later, an optional doorway might reveal the screenwalks, records, and creative decisions behind the fiction.

In this recording, we explore:

  • The transition from research evidence to fictional storytelling
  • How narrative threads and execution branches can inspire stories without becoming the stories themselves
  • A three-act dramatic structure
  • Writing spaces that work like post-it notes throughout the creative process
  • Three possible fictional story hooks
  • The relationship between authorship, human judgment, and AI collaboration
  • A cinematic splash screen imagining what this experience might become

One idea guides the entire exploration:

“I don’t want the robot to write for me. What I want is the robot repeating me, which is different.”

This is a design and research exploration, not a claim that every proposed interface or storytelling feature is already implemented.

Related research

Episode 120 — Plot Threads and Fork Harvest: https://miadi-review-service.vercel.app/review/e29a4fc3-302a-4309-8ad9-39085ee26283

Episode 120 — From Agent to Collective: https://miadi-review-service.vercel.app/review/aa3bb198-59b2-4108-a452-36d25603d5e3

Miadi research issue: https://github.com/jgwill/Miadi/issues/754

What comes next?

Our next exploration could turn the Author’s Desk from a concept into a small working experiment.

Imagine opening a screenwalk, discovering a compelling idea, capturing your own words, and connecting that fragment to a character, scene, or story without losing the research that inspired it.

Then we could explore how a fictional world eventually reveals its own making.

The next story begins with a place to write.

#MiadiFactory #Screenwalk #TheAuthorsDesk #InteractiveStorytelling #CreativeWriting #HumanAI #Chronicle

——

  • Option 1: Expand the Author’s Desk into a tactile experiment where captured notes on ‘A Spark’ become branching character seeds tied to screenwalk traces.
  • Option 2: Introduce a transition layer where ‘A Scene’ fragments merge into an onstage reveal that exposes research threads through optional doorways.
  • Option 3: Use ‘A Story Hook’ to generate three divergent narrative paths that test how agent reflections shape human choices without overriding authorship.

Sources

Bibliographic details of every entry were confirmed by web search on 2026-10-14. Findings quoted above come from abstracts, publisher pages and secondary summaries, not from full texts. The review service stores no transcript for the video, so no claim about its spoken content was checked.