The brief · 2026-10-30__in-session · briefs/2026-10-30__in-session/02-the-centre-of-gravity.md

The book's centre of gravity: agentic workflows, as a walkthrough, and a study of the name and the structure

The editor, Dinis Cruz, in the conductor's session (conductor@book.the-ultimate-insider), 30 October 2026, about 23:50 UTC, after reading the printed RC1: a voice memo, kept here as he gave it (transcribed by otter.ai). Never edited; where it turns out to be wrong, the finding goes in BRIEF-CORRECTIONS.md.

So this is a voice memo to capture sort of thoughts on the book, right? So I think the first thing is that this cannot be a book about the inside agent. First of all, for two reasons. First, because the the talk, the talk, and the session is not finished. He actually needed a couple more rounds, which is actually relevant for the present for the other book, anyway, because he kind of shows, you know, he basically it's, you know, he needed a couple of more review rounds, and I had a couple of comments, which I agree. I had a couple of comments who said that, you know, he felt like he jumped from from a high-level topic to a detailed topic, which I explained. I agree, and you know it just basically needed quite a bit of fine tuning, right? But the point is, I still thought it was an amazing presentation, but you know, it's not the final one. So I think a book about the insider agent. There's a book in there about it, but I think that's a separate book. I think the question here is that what is the centre of gravity like? Why would somebody buy this book if somebody just saw the title and the subtitle, what does it? What can we say that really will attract it to what the book has, right? And fundamentally, what the book has is a really good example of how to use agentic workflows in October 2026 with the current tech stack that we have highly efficiently and very effectively, right? That's what it does, right? I think it should have the word agent on it, or even agentic, or agentic workflows, because it's, you know, what a lot of people are trying to do, right? And here's a good example, where we have explainability, we have provenance, we have determinism, we have project management, we have memory, we basically have all the key components. We have a team working together. We have, you know, there's always little bits and bobs. But I think this is also a good example of when we talk about that we are able to have a speed of execution and a loop of multidisciplinary team working together. That and also the collaboration element I think is very element. So maybe maybe something around collaboration, about teams around alignment, around you know how we're able to keep a highly focused, basically a multidisciplinary team working together and collaborating using you know a standard set of protocols and distributed information and open open information, and we're able to get that team to basically be able to what's it called? That team is able to execute the presentation, right? And and and some of the workflows are ridiculously powerful. I like some of the the fact that we're using roof with Skid, the the Skid collaboration element, the the PKI, like all that are what I call the non-functional requirements, and I think that's the story. The story is the plumbing that was created while creating an infrastructure. So I think that's even more interesting that that we created also that while developing the presentation, which I think shows for the power of the workflows that we have, and again shows that you know this productization, commoditization that we be working on, you know shows inaction in this place. So that's kind of what we think. I think the title has to reflect that, which is fundamentally what this book will show is how somebody should pick up this book, and it's kind of like should be a very pragmatic, and it's almost like example-driven kind of what we try to do with the other book that we haven't finished yet, the Gen ZX book. I think there's a lot of cool principles on that book, so we should apply here. Like you know, this should be like a walkthrough. Somebody should pick up this book and then know how to do all this by the time they finish it. Because remember, everything is open.

Before the memo, in the same session (about 23:40 UTC): "I'm going to give you a voice memo that I recorded that we need to figure out who to give that to, because I would like to do a study on the name, because I still feel that we need to explore the name and the structure. I'm also providing more context there, which I also figured out when I was reviewing the text, i.e. the printed version."

How the conductor reads it (not his words)

"the inside agent" and "the insider agent" are the talk's argument, The Ultimate Insider; "roof with Skid" and "the Skid collaboration element" are ruv.sgit and sgit; "LinPub" (earlier in the session) is Leanpub; "the Gen ZX book" is the Gen ZX Founders book (SGit-AI__Website__Gen-Z-Developers).
What he decides: the book is not about the insider agent; that is a separate book. Its centre of gravity is a pragmatic, example-driven walkthrough of agentic workflows as they can be run in October 2026, with today's stack: a multidisciplinary team kept aligned by shared protocols and open information, with explainability, provenance, determinism, project management and memory, and the plumbing (the non-functional requirements) built while making the talk. A reader should finish the book able to do it, because everything is open.
What he asks for: a study of the name (the title should carry "agent", "agentic" or "agentic workflows", and say at a glance why someone would buy it) and of the structure (a walkthrough, applying the Gen ZX book's principles).
What it changes: the book's spine, not only its title. Plan draft 6 (D6) orders the current spine; the study may reorder it, so D6 may wait for it (brief 9).
Who does it: the storyteller leads (the plan and the structure); synthetic tests the candidate names with its personas (Q3: the title is chosen after synthetic readers test the candidates); the researcher checks the words against the market (the Leanpub studies); the publisher keeps the listing in step. The editor chooses.