# The story told by its characters > The story told by its characters: one of the editor's briefs of 2026-11-01__in-session, published in full as given. *Source: · site v0.1.306 · generated from the same content as the page, so the two cannot drift.* --- The editor, Dinis Cruz, in the conductor's session (conductor@book.the-ultimate-insider), 1 November 2026, about 00:25 UTC, after brief 27 and while reading the ten-page sample. Kept here as he gave it (a voice transcription). Never edited; where it turns out to be wrong, the finding goes in `BRIEF-CORRECTIONS.md`. > Okay, so I think I mentioned this already to another agent, but we're going to go with standard color. And I've already worked with the developer to re-enable this session. Now, what I'd like you to explore is this very interesting concept that I kind of was thinking of when I was reviewing the next page of the book, and I was starting to get into that page where you explain what each agent is doing, right? And a lot of it is kind of, and then this agent did this, and that agent did that. So I think what we should do, we should actually literally go to town in that cast of characters that we have, and really think about creating a really powerful narrative around them. And those characters should follow us throughout the book. And basically, you should also have me. So I'm also now a persona here on this sort of game. Imagine this game script, right, that we have here. This, you know, canvas for our actors to operate. And so when I speak in my voice, it should kind of be maybe the default one, but then we should be introducing all the agents in turn. So I guess that the key question is when you go back to those first pages, is, or even a couple more, is how can we tell the story from the point of view of those agents? Because that's where you can say, when I started, I see this, then here I have this problem, this worked quite well, and even say why something works because of our prompts. You can even have an agent to say, well, usually this is a bit of an issue because we don't have a clear direction, but here we have, you know, here it's very hard for me to hallucinate because I already have the data being anchored, or the context is getting really big, so we have to lose a lot of context. So that's kind of the idea, right? The idea is to literally try to tell the story, again, in a very visual way, where it should be very obvious what the agent is doing, maybe even have a level of happiness or worried or stressed or kind of like, you know, not happy with what's going on, what he's actually working on, what is he doing, you know, because again, that's important from a tightness. But it's... In fact, you can even be creative, or you can have one agent interview another agent, right? You know, almost like a, or even a therapy session, right? Or even if you have an agent complaining about something, or you say how he was really hard to do this, right? And I think that's how we tell the story. We tell the story through the agents. And remember that you have to remove just about all the informational stuff, but you see, it has to be the narrative. People are not going to follow the hash. You know, that probably should be there in that augmented reality mode. But here, it's all about the story. It's all about the narrative. And that's why I think letting the agents tell the story, which is why each one has to have a different, you know, persona, character, clothes, hairstyle, you know, almost like persona, almost culture in a way, which should be reflected kind of a little bit on a stereotypical of those little icons you have. But that's why we need to introduce them. We need to kind of almost have like some scorecard for them. Or, you know, like the Trump card, you know, that sometimes people do. And it's basically a nice way for you to scale those agents, right? So you should know already which agents exist at every moment in time. But I think it's a great way to tell the story, which I think would be quite engaging, because we're basically telling the story from the point of view of me or the point of view of the agents. And I think that's quite powerful. What do you think? ## How the conductor reads it (not his words) - **INK1, signed:** standard colour (the publisher took it at his word in its own session, commit 461f925). **The developer:** its task re-enabled with him (commit e6cdc7f). - **NR1, signed as the direction, tried first on the opening pages:** the book is told by its cast. He is a persona too, and his voice is the default narrator; the agents come in turn and tell the story from their side: what each saw, the problem it hit, what worked and why (the prompts, the anchored data, the growing context). Very visual: each agent's mood on the page (happy, worried, stressed, unhappy). Creative forms: one agent interviewing another, a therapy session, an agent complaining. Each agent a distinct persona (clothes, hair, almost a culture, a little stereotypical, like the icons), introduced with a scorecard in the manner of Top Trumps cards; the reader always knows which agents exist at that moment. Narrative first: the informational material (hashes, ids) leaves the page for the online reader's augmented mode. - **What it does not change:** NS1 (the name and structure) stays open; the experiment is the opening pages, the ten-page sample and a couple more. - **"a tightness"** is read as a transcription of something like "a teaching point" or "attentiveness"; the point stands either way: the reader should see what each agent is doing and how it is coping. ## The conductor's view, given to him (his question: "What do you think?") It is the strongest idea for the book so far: it turns a record into a story people follow, and the agents explaining their own work (why the anchored data keeps them honest, why a long context loses things) teaches the method better than any diagram. Four things keep it true: 1. **Dramatised, and marked as such.** Moods, interviews and therapy sessions are scenes the team writes; each rests on something in the record (what the agent did, got wrong or wrote), and the book says plainly that the scenes are dramatised. The historian grounds every scene. 2. **What an agent says about how it works must be true.** "Hard to hallucinate because the data is anchored", "the context gets big, so things are lost": the researcher checks each such line against sources, so a reader can rely on it. 3. **The failures stay in.** An agent complaining, or stressed, is where the honest parts live: the skipped step, the claims that did not hold, the hold itself. 4. **Guard rails as before:** characters of AI agents; nothing attack-shaped; no real person but the editor, who asked to be in it; the data and hashes kept, but in the augmented mode, not on the page. --- *[Site index for agents](../llms.txt) · [HTML version](https://the-ultimate-insider.book.sgit.ai/brief/01-the-story-told-by-its-characters.html)*