Corey and Christopher Parsons discuss using AI search/synthesis (Chaco) to extract concise, actionable answers from a 400-page book Corey uploaded plus web best-practices. The AI produces project-ready, conversational guidance (examples: how to advocate against vinyl, introduce composting toilets, design for resilience) in Corey’s voice. They highlight that explaining "why" (not just "what") in source text makes it more useful to AI, and emphasize moving from long resources to the right answer for the right person at the right time.
Chapters
What AI search on synthesis can do is not just answer questions like, why is vinyl bad? Or I don't know, whatever else would ask. It can answer questions like, how do I advocate to a client about not using vinyl? Or how do I talk to a client about using. I don't know. I've never gotten compost and toilets into a project. It's always been a goal of mine one day. How do I talk to a client about compost and toilets in their project? Right. And the answers that are coming out based on the content, both in my book and also the content that we've just collected on our Internet that is kind of best practices is unique to us and our values. The answers are really good and they're written in conversational kind of way that a project team can just run with. So how do I make a project more resilient? Like, it's not generic answers, it's kind of explicit, like, do this from a design standpoint. And I asked the kind of AI looking at my book, how do you design for resilience? And it had like six bullet points that are all actionable. And you can imagine a team kind of just taking and using that to evaluate their project. Or how do I talk about resilience to a client? That answer is really good. And the way that I thought I talked about this at a conference last week, the way I talk about it is I've been saying for a long time, we need to get answers, not resources to our project team. We need to get the right answer to the right person at the right time. And then my solution to that was I wrote a book as all the information it needs and the book is 400 pages long. And, and even though I think that 400 pages is extremely succinct for the world of sustainability, it's still a 400 page book. What the, the Corey Bot or like the Chaco AI can do is extract an answer.
Well, you uploaded your book to Chaco, which I don't think we've said.
Oh, sorry, yes, I uploaded the entire book to Chaco.
Yeah.
And. And that's how, that's how Chaco was able to kind of read my writing and deliver an answer in my voice with information that, that I've provided.
Yeah. And alongside the Internet, which I think. So there's two things there. Like one, you wrote a lot of this stuff pre AI, but the way that you develop content was very like AI ready in that you didn't just save happy accident, you didn't just by happy accident. Well, I don't know. You tell me, I'll tell you how. What I meant tell you. So, like, a lot of people focus on the what when they're writing kind of knowledge internally, right? So our window to wall ratio is between 30 and 50%, full stop. Like, that's the end of it. Not, like, why is it between 30 and 50%? Like, when would it be 30, when would it be 50? Like, why is this our approach? And when you look at any of the content on Bora's intranet, the why is always involved, like in the who. And like, there's a lot more context there, which is great for humans, which is why you wrote it that way. But it turns out it's great for AI too.
So I think you're right. And I wrote it for. I wrote it for. I did. I wrote it for people, and you.
Wrote it for people.
And there's this, this joke that I've made in the past where it's like, whenever there's a question, sustainability, the worst answer and also the most accurate answer is it depends. And you hear that a lot, right? And it's like, it depends. It's always true. Like, what's the. What's the best window wall ratio for this project? Like, well, it depends on the context, depends on the orientation, depends on the building type. But it's completely unhelpful to say what it depends on. So that's why it's the worst answer.
But you're saying why it depends. So you are actually like. And then you go on to like.
So what I tried to do in the book was like, explain at a high level. Like, not going to every single level of detail, but at a relatively high level. This is the best answer. This is why it's the best answer. These are maybe one or two use cases where it might be a little different. And this is how you know. But that's it, right? It doesn't go in through every single nuance, it doesn't go on forever of every single possibility, but it provides enough context that the reader should be able to intuit kind of the next step. And I think you're right. I think that's what allows the AI to pick up on that explanation and make it maybe more accessible with no intention at all towards writing for an.
AI without intention, right? But people can learn from that. And now they can future proof or AI proof their content to make sure it's more.
