The brief · 2026-10-31__rc1-paper-review · briefs/2026-10-31__rc1-paper-review/batch-09-pages-39-to-44.md

Batch 9: pages 39 to 44: context before detail, no judgement of others, screenshots, the cast, ship end to end

The editor, in the conductor's session, 31 October 2026. Kept as he gave it.

Now we're on page 39. And again, like, you know, if you look at all of the information, the talk throws, the five stack, the editor returns, all of this, none of this has context for the user at this stage. In fact, most of this would have much better be described when we get to it with an infographic, right? With a sequence of events, with like the little micro animations or micro explanations, right? Like how to tell this in a visual way, because, you know, there's no context. And until the users have context, they don't understand this, right? And in the on page 40 is the whole, actually, I don't have a problem because I don't have page numbers, right? This is interesting, right? I think I should be able to hopefully connect them from here. I got also it open on my iPad, so at least I should be able to connect the dots, right? So yeah, okay, so that's 40, where the breaks sit, right? So where the breaks sit, that's a good example. I need to rewrite this, right? Because it's a good idea, but, you know, I think I spoke enough about why this is needs to be improved, right? It's not just breaks, it's the whole technology that works around it, right? And then on 41, you have where this come from. Again, massive stuff that you don't need to fight to care about at that moment in time. Okay, so now we go on 43 page, which is midnight, an empty repository. I don't know, that I really don't like when we say things like this, like in that order and why the order matters more than any single thing on it, right? Like a lot of this stuff, I always find that it passes judgment. we say, oh, we're going to talk about it because nobody else is talking about it. It's literally a lose-lose, because either the other person's not talking about it and we just, you know, slack someone off, or, you know, and basically like demean somebody, or they're doing it, it's obvious and they know, well, this is wrong because I know at least of one case that is actually doing it. So let's not shoot ourselves in the foot, right? And not commit. Now, also on page 43, again, like you talk about, you know, one script, a gate, release pipeline, version number, talk to this, a keyboard player, blah, blah, blah. All of that, we need screenshots for this. First of all, we need screenshots of actually what it looks like, and then we need infographics visually. Talk about it, right? So now on page 44, so when you talk about feet of speeds or shape, blah, blah, blah, this is the first paragraph. I think the thing about what's the power here is that CI pipeline. That's what allows us to have the level of speed. That's what allows us to implement this. By the way, that's one of the things about the LLM-driven development, is that we can basically use the LLMs to help us a lot with this sort of maintenance and a lot of these capabilities, right? Now, when we have the brief, the MD, and reality plan MD, now that we have more width, right, it's super important that you always show a screenshot of parts of the page or a little section of it, right? And maybe put a screenshot on the left, on the right, and then the text on the left, right? But we need to show examples. We need to bring this to reality. At the moment, it's just brief MD, which is a bunch of text. I have a reality MD, a bunch of text. I think these need infographics, need images. And then you explain it. But you need that visual representation of what it is and also where it's used, right? Like, who uses this, right? So we need to again say, is this a file for the agent? Is it a file for the sub-agent? Is it a file from the consumer? Like, you know, again, which policy governs this, etc., right? And actually, for example, I think here, this is probably a big thing now, right? I feel that, you know, you need to think of this as a movie and as a sort of a cast of characters. So I think that one of the first things we need to do is introduce the cast of characters, right? Like, who are they? What are they like? What are they doing? You know, what's their focus, right? And present that again in a visual way, so we don't write, you know, 5 pages about it, but then we say we have all these details, maybe show detail for one, right? Which is kind of why I'm thinking that maybe the way to do is you start with the conductor, right? The conductor is the first one. And we say, look, this is the one that has the job of coordinate everything, right? And some people might want to talk about, you know, be a project manager or be this thing or be that, but, you know, I think in our cases, the logistics is very important, right? So without that, none of that thing works. So conductor, the CI pipeline, all that stuff, which of course now we provide a lot of it because we already have a lot of those examples open source, but you have to start there. You have to start by shipping something. And this is where I said to the first time of the team, look, just get a PDF out, right? Just do something end to end so that we have, and like even if you look at the presentations, one of the first objective was just create a deck. Let's have a deck and then we iterate on it, right? And then put a CI, put a visualization element to it, improve the UI. But at any moment in time, we were producing something that was shippable. And I think that is super critical. And so show the example of the files. And in a way that's, you know, show the part of the CI pipeline, which again, you guys should be experiencing left, right, and center, right? I think one point I'm not sure I made is that we can use the examples, which are, I guess, are meta examples of this, you know, this own team that we have here, right? It's a great example of this working. You know, the team that we set up to do the code read, that's another cool example. And that might be a simple example to explain, hey, this is how that works. You have these agents, you have that, you're running Claude Code. And it's, again, all about providing the details for the user to be able to replicate this. And again, we will then get our synthetic users to also replicate it and provide information about, you know, maybe where you get stuck and what you needed, etc. Which again, with that augmented reality on the side, allows us to do this very effectively. Which is also why we then need to have this very automaticity, right? You know, like this very sort of, is this very sort of packaged and self-contained deliveries, right? And self-contained, let's say, books in this case, where, you know, all of it should live under a folder, right? All of it should live in a couple of folds, but ideally in Git, because, you know, it's a little bit more stable, I would say, right now. But all of those materials get saved there, right? And I think that's super powerful. Now, yeah.