206: ‘Purpose-Built AI: Why It Matters in AEC’, with Rachelle Ray

206: ‘Purpose-Built AI: Why It Matters in AEC’, with Rachelle Ray

Evan Troxel: Welcome to the TRXL Podcast. I'm Evan Troxel. In this episode, I welcome Rachelle Ray. Rachelle is a veteran, AEC marketer, who spent more than 15 years inside firms and as a consultant, and today she's helping build purpose-built AI at OpenAsset, specifically Shred.ai, a proposal tool designed for the realities of our industry.

If you've ever fought your way through DAMs, resumes, project sheets, and "final_final_v5.pdf", you'll appreciate what she's working on. In today's conversation, we explore practical approaches to re-imagining proposals with AI, addressing the persistent big three challenges, which are content management, collaboration with busy subject matter experts, and incorporating meaningful client insights into deliverables.

We examine how domain specific tools are transforming possibilities in the AEC industry, And our discussion covers AI procurement strategies, the nuances of writing for both humans and machines, why human oversight remains essential, and the significant time savings achieved by new tools that get us to first drafts much faster.

We also explore governance and trust issues, including compliance considerations, information segmentation, and why leveraging your firm's proprietary knowledge consistently outperforms generic prompts.

A key topic from this conversation is speed, but the real story isn't about doing more proposals faster. It's about what leaders choose to do with the time that's returned. Do you reinvest it into strategy, storytelling, and client alignment, or do you squander it on volume? Today we talk about how firms are using this freed up time to unlock insights from their SMEs, deepen client research and ultimately produce more compelling proposals that align with the kind of work their people actually want to do.

so now without further ado, I bring you my conversation with Rachelle Ray.

Rachelle, welcome to the podcast. Great to have you.

Rachelle Ray: Thanks, Evan. Excited to be here.

Evan Troxel: I'm happy to host you and, and we're gonna be talking a lot about marketing proposals and AI and kind of what firms are dealing with with all this.

But before we get into that, maybe you can give us a bit of your background.

Rachelle Ray: Yeah, my background encompasses all of those things. Um, so I've been an AEC marketer for over 15 years. It's been a while. Um, I started kind of when everybody was moving from PageMaker to InDesign to, uh, you know, date myself there.

Evan Troxel: Yes, me too.

Rachelle Ray: Super fun. So I've seen technological shifts happen, right? Um.

Evan Troxel: You, you skipped Cork Express, though. You didn't mention Cork. I don't know if, I don't know if that was one of your,

Rachelle Ray: Uh, so they had just stopped teaching it in school when I got there,

Evan Troxel: all right.

Rachelle Ray: I'm aware of it, but did not get to use it, unfortunately. But

Evan Troxel: All right.

Rachelle Ray: PageMaker was fun. Oh, no, no, you're good. You're good. Um, I worked in-house for large and small firms and then I've been consulting for the last eight years for AE and C and everything in between, all of the specialties.

So I have like a 360 view on the industry, which is really exciting and really interesting. Um, but I had this big shakeup in 2023 in the fall. Um, this is when OpenAI threw chat GBT out into the world for public access for the first time.

Evan Troxel: Mm-hmm.

Rachelle Ray: And that was my first experience with it because I, you know, kind of had my head in the sand on this whole AI thing.

I didn't know it was coming. Um, so big shock for me when I'm playing with this little chat bot and realizing. That, uh, it's not great in some respects. I do remember getting returns like I'm just an LLM, I can't help you with that, which was super fun. It doesn't do that anymore. Um, I wish I had screenshots, but it was also starting to return really good like LinkedIn posts or blog posts.

And I kind of had this career freak out where I realized there is a possible world where AI writes as well, if not better than I do, in which case, what the heck do I do? And so I had this career crisis where I debated going back to school for engineering, remember that I hate math. Remember that computers are really good at math and um, so maybe that's not the best place to hide from the robots.

Evan Troxel: engineering? What, what, what was it about engineering? Was it, was it like co coding and things or, or some mechanical

Rachelle Ray: Like mechanical or electrical engineering because I really enjoy working. Probably civil, uh, I don't know why I am a weirdo. Civil engineering proposals are my favorite. I love transportation engineering. Um, so I was kind of eyeing that and then went, Nope. Computers do calculations really well. They're coming for them too.

So instead, I decided there was nowhere to hide from the robots. I should just get really good at working with them. And so that's what I did. I started really doubling down on how I was using this technology. I started playing with as many tools as I could get my hands on. I started sharing, um, thoughts in the community that I run about, like, this is how I'm prompting.

How are you prompting? This is what I'm doing. What are you doing? And so now I've kind of come to this new career path where I'm taking ownership of the AI in the marketing, um, and teaching other people to do it as well. And looking at, okay, we are still marketing. It's not gonna replace us. I now understand the limits, but it's gonna enhance us and it's gonna change how we work, how we think, and so how do I merge those together?

And that's how I ended up at OpenAsset, helping to build our purpose-built AI proposal tool and bring those three things together.

Evan Troxel: I am sure OpenAsset is a familiar brand with a lot of our listeners because I, I think a lot of our listeners here are in larger firms that employ such platforms. Could you just give a quick overview for those who don't know what OpenAsset is to just tell the world about it?

Rachelle Ray: Of course, uh, OpenAsset is a digital asset management, uh, platform. So it's basically where you put all of your project information, your people information, all of your gorgeous photography,

Evan Troxel: Hmm.

Rachelle Ray: data, your employee resumes in one single source of truth that's easily searchable. And so it makes it very quick to access the assets that you need for marketing, uh, daily.

Evan Troxel: Okay. And, and when AI. Came to OpenAsset. In what way did it come in? I, I'm trying to remember if, so, correct me if I'm wrong, but I think it was 2018. 2019, I was at a large firm round table meeting in Portland, Oregon where OpenAsset came in and did a presentation. And I believe that was the first time I had heard about AI in OpenAsset.

And it was really around image search. It was kind of like, what if I needed to find all the projects that had brick in the photographs? Right. And, and

Rachelle Ray: Yes.

Evan Troxel: like returning results based on that. Does that sound familiar as far as like an early use of AI at OpenAsset?

Rachelle Ray: Yeah. Yeah. That was one of the first use cases that we kind of looked at was, okay, you know, I want all of my. Bridges at sunset, right? That's something that we would hope that AI would be able to do. And so we absolutely do have that AI search now. Um, it's really good at finding, you know, materiality, like you said, brick walls, glass facades, stuff like that.

Um, it's also integrated into the written part, you know, if you're working with your project description or somebody's bio and you need to make it shorter, longer, expand on it, it looks at the information it already has, helps you build out that information, that bio or whatever that writeup looks like based on content it already has.

So it's not like hallucinating or making anything up. And those are just kind of the, the starting points of where we are with OpenAsset. We have this beautiful, amazing vision for what the DAM side will look like when we have, you know, this more advanced AI search to look through all of your assets.

And of course, you know, we've got Shred.ai on the other side that's gonna be, um, super leveraging AI to help you with proposal management and proposal development specifically.

Evan Troxel: Okay, so marketing in AEC firms, this is kind of a longstanding thing, at least on this podcast. I've had previous guests that we've talked about this. Marketing in AEC usually means. writing for RFPs. Right.

Rachelle Ray: Yes.

Evan Troxel: and, and so, you know, there's, there's marketing the firm, which is, could be seen as, you know, putting out what, what are your distinguishing attributes, what are your values?

Things like that. But I think a lot of times our marketing departments are specifically aimed at responding to RFPs. And so development from that angle is, is pretty typical way that that firms are going after work. And that's what their marketing department's kind of built around. So what keeps you really close to what's happening in AEC and business development when it comes to, you know, your daily, your day to day?

Rachelle Ray: Yeah. Um, it was a provision of me joining OpenAsset that I get to stay enmeshed in AEC marketing. Um, I can't let it go, so I still run my consulting business. I have a small team. Um, I don't actively write a lot of the proposals, but I'm still QAing them. I'm still problem solving with them. We're still, they throw things at me.

If they get a new RFP requirement, they're like, what the heck is this? I've never seen this in a proposal before. I get to troubleshoot stuff like that and then, you know, pass it along to the team. I run a proposal community as well. It's not AEC specific, but it is proposal specific. And so I'm seeing insights from what's going on in the tech, uh, industry, in healthcare, um, in accounting and all over.

Um, and usually what's happening in tech will eventually trickle down to AEC. It might take a little bit, but I can forecast based on what's going on in, you know, more technologically advanced. Sorry, AEC. We're a little bit behind the times, but more advanced industries.

Evan Troxel: person to say that. Yeah.

Rachelle Ray: We're a little archaic. We all know it, but it's comfortable, right?

But the point is that I can, I can look at these more forward thinking, more, um, tech, uh, forward industries, see what's happening in their procurement, and then start forecasting what might be happening in AEC in the next 2, 3, 10 years type thing, which is really cool as well.

Evan Troxel: I didn't see that coming. So can you give any insights on what, what kinds of forecasts that might be on the horizon? I mean, this is crystal ball a little bit, but like you say, you're seeing this happen in the tech side, so any, any hints is what those could be.

Rachelle Ray: The biggest one that everyone's talking about right now is AI procurement. So if procurement is using an AI to evaluate your proposal, not a human.

Evan Troxel: Yeah.

Rachelle Ray: does that change how we write our proposals? Um, for tech, it's a lot of like q and a pairs, right? It's text to text. But AEC is a very visual industry, so trying to figure out how that's gonna change for us has been really interesting.

Like Will and AI be able to read an architecture proposal the same way it reads like an ed tech proposal?

Evan Troxel: to, to hear that that's coming, but it's

Rachelle Ray: Yes.

Evan Troxel: I mean, this is how many, like, I think about HR and I think about, you know, job applicants and reviewing and kind of whittling down the list of thousands of applicants to try to get down to the handful that should actually be interviewed.

But, and so you're, what you're saying is like getting ahead of that curve to write the proposals, knowing that that's how they're gonna probably be vetted. Right.

Rachelle Ray: Right. Right. I wouldn't be super doom and gloom. I think AEC is still very much a relationship built industry, and so we'll have an opportunity to shift towards more, you know, maybe, maybe we hide from the robots in capture planning and the business development side, and we really invest in those relationships so that the human on the client side can maybe circumvent that AI and make sure that you do get to the second round of the human evaluation or whatever that is.

I think that might be a shift that we could see that will be unique to us.

Evan Troxel: You, you, you may, you maybe didn't use this word, but you mentioned something regarding like curve balls that get thrown at firms in the RFP response

Rachelle Ray: Yes.

Evan Troxel: process. what are the kinds of challenges? Because I, I imagine like, it's probably 80 20, right? Like 80% of of the RFP response is pretty much, you've done it before.

You've probably done this before, and you get to kind of almost copy paste with a little bit of Right specif specificity for, for this, this response. But what are the kinds of challenges that AEC marketers are, are telling you that they're seeing and, and they're needing, you know, a little bit more insight on.

Rachelle Ray: So this is, this is very disheartening. Um, but since I started and where I am now, we have faced the same three challenges as AEC marketers just across the board. I hear it from people all the time. They literally will come up and tell me this. When I attend conferences, they'll shout me out on LinkedIn and say, I'm still dealing with this.

Um, OpenAsset has surveyed our customers a few times and we keep seeing the same thing. And that's three challenges. Number one, we are still struggling with content management. Where do we find it? Where do we store it? How do we access it faster?

Evan Troxel: mm.

Rachelle Ray: Number two, collaboration. How on earth do I get my technical staff, my architects.

That are on their billable projects to pay attention to my proposal and give me the brilliant insights that only live in their brains.

Evan Troxel: Yep.

Rachelle Ray: number three is getting the business development insights, the client insights into our proposals so that we can do that. You know, 20% tweaking because that is the most important piece of the proposal, right?

So same problems, still dealing with the same misconceptions of, oh, you don't actually need me because marketing is easy. Proposals are easy. I think you just said it, Evan, right? Like you just go grab that boiler plate. You go grab that pass, go by, you copy paste, you spruce it up a little bit, you send it off.

But that's really not what we do.

Evan Troxel: yeah. Right. A, as an architect who is, you know, like you said, it's like I'm working on the projects I have today and,

Rachelle Ray: Right.

Evan Troxel: team is focused on the projects that are next. Hopefully like building the pipeline, building the backlog, and it is easy to, to say like, I don't have time for this.

Right. To, because I have so much to do right now. Man, it, it's a tough position to be in as like a marketing manager in an AEC firm because you're constantly hounding, I'll use these words to kind of describe hounding chasing, right? It's like putting out the smoke signals, anything. Please just respond to my message and, and give me some time so that we can actually do something a little more meaningful, right.

And get the good stories, because you're right, the stories are locked away in their, in their brains. And it's like, those are what you actually want, right? Is like the,

Rachelle Ray: Yes.

Evan Troxel: stories, the, the interesting challenges that were solved and those rarely get written down made available to anybody, right?

Like,

Rachelle Ray: Exactly.

Evan Troxel: Yeah. That, that is a tricky position to be in. I, I'm, I'm cur, maybe we can come back to this, but I'm curious like if you've seen any success in like how to actually, how people have been able to do that.

Rachelle Ray: Oh, yes,

Evan Troxel: little bit about it on the podcast with, when it comes to like knowledge management and interviewing and just making things like super low key. And just approachable when it comes to, you know, this isn't a formal thing. You don't have to list it, you don't have to write it out. You don't have to think, like, just tell. So, so

Rachelle Ray: yes.

Evan Troxel: to some, some things that, that you see as like successful ways to kind of unlock those stories. But I think that

Rachelle Ray: Oh, for sure.

Evan Troxel: of that kind of stuff is, is, you know, this is a softer side of, you know, these are diff different skills than it comes to, like the technical challenges of serving, you know, an architectural services need.

So, a lot of people maybe just aren't prepared to, to take the time to do that. But, I'm, I'm curious, like when you talk about this copy paste and we talk about boilerplate and, and, and then we're, we're adding in AI to that, to maybe, maybe that's doing some of the sprucing up, but I'm, I'm curious how you're seeing AI when it comes to what, what you guys are working on. How are you using it to enhance, you know, what's already there? Is there, are there things that you're seeing that are. Really like, whoa, I didn't expect it to come up with with that, but because, because there's enough information in there for it to pull from and maybe connect some dots that haven't been connected before.

How, what, what's playing out in that realm?

Rachelle Ray: My favorite thing about AI is like there's no one right way to do it, and there's no like one prescribed way to do it. And so I'm seeing all of these like cool different use cases and I just kind of like. Like a little dragon. I just hoard all the best ones.

Evan Troxel: Is

Rachelle Ray: Um,

Evan Troxel: do I

Rachelle Ray: they hoard gold, right? Like, I don't know,

Evan Troxel: giant pile of gold. Yes.

Rachelle Ray: a giant pile of golden insights.

That was, that was how I envisioned this, you know? Yeah. Like smog style.

Evan Troxel: Right. Okay.

Rachelle Ray: Um, but so there are a bunch of different places where I'm seeing marketers using AI from, you know, before the proposal you even pick up your pencil and write the first thing or hit the keyboard, first sentence. They're using AI to research the client, research the agency, figure out what their challenges are, who the people are, understanding the selection committee as human beings and individuals.

And then in the broader scope of like who, how they fit into the institution or the organization or whoever it is, which is really cool. 'cause then you can take those. And weave them in with your boilerplate. And then you don't have boilerplate anymore. You have these like super curated content blocks based on that information.

Evan Troxel: a how question there?

Rachelle Ray: Yeah.

Evan Troxel: if, if I was gonna approach that, it seems like a great. Use case for kind of a deep research, kind of a model in where, and, and maybe you, you add in some specific links to like LinkedIn profiles or about pages on websites so that it, it really prioritizes those.

But I would imagine there's, there's other stuff that's pulling from too, but that seems like a good way to go about it. What, what do you think?

Rachelle Ray: I've tried a couple of different ways. Um, perplexity is really great for kind of like a,

Evan Troxel: Yeah.

Rachelle Ray: a nice balance between deep research, which can take a really long time if you've run one of those. Sometimes you can like go get a coffee and then come back and it's still running

Evan Troxel: A really long time. Yeah.

Rachelle Ray: a really,

Evan Troxel: 10 minutes. Yeah,

Rachelle Ray: in AI land, like 10 to 30 minutes is a long time.

Um.

Evan Troxel: to what, what? Remember how Google, and maybe it still does this, but because I haven't, like, okay. I just, I'm not using Google that much anymore, but it's like, remember, it would always say like, oh, it returned these results in 0.006 seconds or

Rachelle Ray: Yes,

Evan Troxel: we got trained by that for sure. And so now you're right, like 10 minutes is forever in ALA to re to respond to a query.

Rachelle Ray: exactly, exactly. So perplexity, you know, does it in 30 seconds, which compared to your 10 minutes is, is much faster. It provides all of your links and such. Um, I have seen people use like dedicated business development ais like crystal. You can just attach as a, um, like an add-on to your browser. It looks at LinkedIn profiles, pulls up like your, I think it does a disc assessment, like personality assessment.

And you can just grab those insights, you can extract them and then. Like layer them on to chat, GPT, be like, here's the selection committee. Here's the information I found out about them. Dump all that in and then start interacting with it. Um, it's a lot of blending, I think of tools.

Evan Troxel: I was gonna open up perplexity right now just to look at

Rachelle Ray: I love it.

Evan Troxel: um, one of the things that I think is kind of an interesting way to go about, so solving this problem too, and maybe it's not like focused on the, the who of, of, you know, the, the organization that you're to for an RFP, but like what's the problem they're trying to solve

Rachelle Ray: Right?

Evan Troxel: a really interesting thing to think about when you're, when you're responding to an RFP.

And one of the tools in the perplexity research is that you can actually focus it in on what they just label as social, which really just means Reddit. Reddit is one of those places on the internet where it literally is kind of the forum of the internet.

Rachelle Ray: Right.

Evan Troxel: you can ask. Perplexity to do deep research into social, to actually talk, like, think through what could the possible challenges be, what could the possible solutions be from a very, like, like it's super, you know, these are one-off comments or comment threads, and it's really real.

Like when it comes to

Rachelle Ray: Yeah.

Evan Troxel: who are writing that stuff, there's no filter, right? And so, um, I think Reddit's pretty well known for being a, a no filter kind of a place. So you get, think, more honest and and authentic, um, thinking happening on there because it's so off the cuff and, and there's not a lot of, it's not a blog post.

It's not something that you're, you, that needs to stand the test of time. It's very much like, just a, a co the water cooler talk of the internet. And I could see that as, you know, one of perplexity strengths is you can actually narrow down the research to just work in the social aspect of the internet, which is kind of cool.

Rachelle Ray: It is. And I, I would say merge that. Um, when I run a deep research with Chachi pt, I intentionally tell it not to look at social because I want like the, the published quote unquote, um, true stuff. I know, I know

Evan Troxel: Yeah,

Rachelle Ray: that in quotes.

Evan Troxel: using your famous podcast, air quotes. I get it.

Rachelle Ray: Um,

Evan Troxel: time,

Rachelle Ray: but you merge those, right? So like you have the official stance, uh, deep researched and then yeah, you would have your social stance researched and then you're merging. What is the institution saying? What is their stance? What are their people staying and is there, is there an alignment there? Is there a discrepancy there?

And that's creating a challenge. And yeah, then you get those like beautiful little nuggets that you get to stick in your proposal.

Evan Troxel: Yeah. Yeah. Very cool. I'd like that you were able to share some tips there, because I think some people, you know, usually it just touches the surface and says, well, we're using AI to do this, but this is actually, I think, a little bit more useful information for people to give it a try and, and. It's really interesting in the research. I mean, this is not an ad for perplexity, right? But it, a tool that weaves together these different aspects. So you can like build a perplexity thread out of just researching social, just researching the web, and then you can tell it to weave those together just by clicking a couple of different buttons in the moment of when you're actually running it.

So

Rachelle Ray: Right,

Evan Troxel: quite useful in that regard. Yeah.

Rachelle Ray: exactly. Yeah. Leverage everything you've got.

Evan Troxel: yeah, I, I agree. Leverage everything that you've got. Um, and, and because it gives you a little bit, I think, I think this actually frees up your brain to to look for those nuggets a little bit more because these deadlines are crazy, right?

When it comes to responding to these RFPs, like, these people are running a mile a minute and, and it's like, this actually gives you a little bit of extra time to step back and think a little bit

Rachelle Ray: Right,

Evan Troxel: about. of these new, fresh, you know, perspectives that could, it could potentially bring to surface and then you can choose which one you wanna go down.

Rachelle Ray: right. Exactly.

Evan Troxel: So is there any other ways in which firms are kind of using these kinds of tools to overcome friction in the process that you can think of? That are in addition to like just uns surfacing or surfacing information inside their own firm?

Rachelle Ray: Exactly. Um, that was gonna be the next, you know, we did before and now we're during the proposal. Uh, you have so many amazing insights in old proposals, in thought leadership and award submittals in your content libraries. We have a million sources of content and information that we need to get to, to put into our proposals.

And the bigger your team gets, the more siloed that often becomes,

Evan Troxel: Hmm.

Rachelle Ray: you know, I don't always have insight into the proposals my team's working on. Like, I'm, I'm aware. That someone is working on a proposal for the university or the city or whoever it is. But I don't know the stories that are going into that proposal, right?

Because I'm not the one actively working on it.

Evan Troxel: Right.

Rachelle Ray: And so AI being able to look at our entire, like body of knowledge for our firm and see all of those stories because it has access to everything. It can now surface those insights for us. It can take those challenges that we've identified or those individual tidbits of information that we wanna focus on, and we can say, okay, go forth and find us, you know, every really incredible sustainability story for our lead platinum, you know, projects that have done or been accomplished in the last five years or whatever.

And AI will just crawl through and grab that. So I don't have to do this manual search anymore, like Del Tech, where I'm like clicking all of my options or even, um, an OpenAsset, like you have to filter it. AI will just scour return and say, did I do a good job? And if you say, yes, I'm personifying, engineering will kill me for that.

Uh,

Evan Troxel: It is still on. It's still on you though, to check, right? Did

Rachelle Ray: it is

Evan Troxel: Well, I guess we need to find out,

Rachelle Ray: yes. Human in the loop.

Evan Troxel: sure.

Rachelle Ray: Exactly.

Evan Troxel: right.

Rachelle Ray: Yeah.

Evan Troxel: you're, you're now talking about kind of a more integrated, rather than like this bolt-on approach of using

Rachelle Ray: Yes.

Evan Troxel: tool. So an inside tool inside the firm using the firm's information versus something that only has access to what it can see that's out on the internet.

Those are two very different things, and so there is some additional in that. you just kind of talk about like what. How, how did, how is that accomplished? So does the firm already have to have all that information in OpenAsset? Do you link up to other things via APIs? Like how, how is that working inside the firm?

That, that, I, I hope would take a little bit of that pain away as well, because that's a daunting task because that super administrative, you, you think about

Rachelle Ray: Yes.

Evan Troxel: come in to do over the summer. It's like, do data entry. Right? It's

Rachelle Ray: Right,

Evan Troxel: need it. And that, that doesn't sound too enticing for an architectural intern.

Maybe it's, maybe it's a marketing intern, I don't know. But, but those kinds of things are, are real struggles that firms have to deal with as well. So how is this actually working inside firms?

Rachelle Ray: for sure. Um, as a marketing intern, didn't love it either. Just saying, um, I can't really speak to other tools, but what OpenAssets, um, Shred.ai will do is create a content library where you can just drop, like drag and drop your proposals into it, your final PDFs, your thought leadership, and then you interact with it there from the platform.

Um, we are looking at, you know, potential options for pointing shred, uh, Shred.ai at your, you know, SharePoint or whatever it is to make it even easier for you. We really want to meet marketers where they are. We understand that tech stacks are exploding, like I said, you know, grabbing all of the tools that you can, but that just, that just, we get technological fatigue from that.

And so we're trying to, you know, figure out how we can deal with that. But for right now, you have your library, it's like a digital shelf. You dump your information in that digital shelf and then our AI takes off and ingests that information and then you interact with it that way. Um, almost like setting up a, um, project or an agent in Chachi, bt, or Claude, or a gem in Gemini.

Evan Troxel: I, questions come to mind thinking about that. So yeah, like project data, living in folders on the network is basically completely opaque, right? It's like, I, job

Rachelle Ray: Yes.

Evan Troxel: is it? What marketing segment is it in? And it's like, okay, well that's a job just to find where to go find the information.

Rachelle Ray: Yes.

Evan Troxel: difficult. Um, and so what you're saying, I guess, is like, take all those PDFs that have been created that are kind of these static documents that say the final final five B, no, really this time final, this is the one that went out to the client, right? We all have those

Rachelle Ray: Yes. That was too real.

Evan Troxel: in there.

Uh, take the last ones, dump 'em in there, and so then it's gonna catalog those. And now, and so that brings up another question and, and I mean, this is just more of like throwing, throwing a bunch of information across and seeing how you respond.

Rachelle Ray: Yeah.

Evan Troxel: One thing that I've seen with LLM working with LLM specifically is like, is really good at looking at the beginning of the prompt if you're using long prompts and it's really good at looking at the end of the prompt, but in the middle not so much.

Right. And so like, it's like one of the strategies when it comes to working with LLMs is do it in steps, do smaller steps, and then reinforce, go back and say, okay, now using that information. And so I'm, I'm curious if you dump a bunch of information in there from these kind of done PDFs, which are not small documents, right?

Rachelle Ray: No,

Evan Troxel: often, you know, a hundred pages more, I don't know. And, and so how well is it digesting that information to surface information?

Rachelle Ray: for sure. Um, ais are a little bit lazy. Uh, that's probably not the technical term for it, but there have been a lot of studies on this and like whether they're saving computing power or if there's something going on with the prompting. But the way we are approaching this is to kind of break down the ask of the ai, so really directing it towards what it needs to look at the most, right?

You wouldn't approach your library and just say, tell me everything, right? You're usually on some sort of a mission as a marketer, like, I need to know all about our. Pursuits with the city of Atlanta since 2022. Tell me what we've done there. Right? Or gimme all the storm water projects that we've done in Daytona, right?

And so we're directing it a little bit. Um, it is chunking the information to understand it. That's just a process of breaking it down into smaller bits of information that it can ingest and digest a little bit better before it spits an answer back out at you. Um, so we use, uh, I'm not even gonna pretend to be able to explain RAG and data entry and retrieval and all that.

Um, I know RAG is, you know, retrieval, augmented generation. Um, so really focusing on the r the retrieval part, making it as easy as possible and as clear as possible for the AI so that it can do a good job and go get the information that you're looking for. And you're not just like sending the intern off to guess.

You know of this, of this big, broad, giant library, where should it go? We're trying to put the guardrails in place to direct it to the most appropriate spot in that library to focus its efforts and then return even better responses. And if it can't do that, to come back to you as the user and say, can you clarify?

Right? Like building that fail safe into the AI to help you help it. Kind of like, kind of like you would with an intern if they went off to do an assignment, didn't understand the assignment, they come back to you.

Evan Troxel: You would hope. Yeah.

Rachelle Ray: You, you would hope.

Evan Troxel: I, I'm a father, I have kids. They don't do that. Um, yeah. Anyway, um, I mean that, that, that kind of puts us into the, a good segue into like keeping people in the loop of this whole system. You mentioned it a minute ago as well regarding, you know, just, know, chunking by, by doing this kind of force chunking, like I give it a thing and then I have it give me, and then I give it another thing and then it responds and I give it another thing that keeps me in the loop and make sure that we're going down the path that I want to go down. so, I mean, thinking about. know, you mentioned it early on about like it's the robots are coming for, for my job and should I consider doing something completely different? I think a lot of people are, are, are think thinking, like, is that just a myth though? Do you really feel like this is, you know, maybe that's a myth.

Maybe there's other ones that, that we could just dispel early on in the conversation around like at least what you, you think is coming versus where we can make a real impact.

Rachelle Ray: I, maybe I'm optimistic. Um, I'm gonna borrow a term from somebody else who says they're a techno optimist. I'm gonna, I'm gonna take that. Uh, I don't think that the robots are gonna take our job. I don't think AI is going to replace humans. I think the bigger fear is humans not understanding that technology are going to replace humans with ai and then realize that that was a bad idea.

We've already seen that happen in, um, tech proposals. We saw.

Evan Troxel: I think in headlines right now, too.

Rachelle Ray: Yes,

Evan Troxel: talk about how, you know, CEOs are getting rid of staff and they're, they're boasting about it and, and they're replacing it with ai and, and I kind of wonder if there's a backlash coming from that.

Rachelle Ray: we've already seen that happen. Um, you know. We laid off a bunch of journalists and blog writers and content creation professionals in 20 23, 20 24. Uh, realized that content wasn't great. That was being churned out by the ais, hired a bunch of people back. We saw it in proposals too. These automation tools.

The early ones came out. We saw a ton of people getting laid off in proposals. Uh, the tech tools didn't do exactly what they were supposed to do because you still need a human in the loop looking at one.

Evan Troxel: Right. Not just a, not any human. Someone who knows.

Rachelle Ray: Someone who understands, yeah, you don't want me, I tell people all the time, you don't want me over overseeing the engineering ai, and I don't want you the architect overseeing my proposal.

Ai. So you need somebody who knows, like is it returning good outputs on how it's pulling apart or shredding an RFP? Did it get the client name right? Did it get the response criteria right? AEC proposals are notorious for having. Like a throwaway inclusion, right? Oh, don't forget, we want your audited financials for the last 17 years.

Um, not in our response criteria. It's just like casually tossed in somewhere and you're like, oh, okay, cool. Let me go find that. Or we need eight references for every single person that you've listed on your team, and that's also just casually thrown in somewhere. So is the AI catching all of that information, right?

Is it doing it well? Is it fully understanding the RFP? And if it does, is it then taking that information that understands to build an appropriate outline, to build an appropriate draft? Did it see that you needed to have on-call experience and go grab all of your engineering on-call write-ups instead of the architecture ones?

'cause it's for an architecture proposal. Like, did it understand? And you need a proposal professional. You need a marketing professional to see that, to guide it, to tell it, Hey ai, you did a great job here. Or a, oh no, you missed the mark completely. Here, let's try this again and guide it. So I don't think we're getting replaced anytime soon.

I've seen enough of the AI doing some really weird stuff, um, to not trust it to steal my job, but I think it's gonna change us. I think we're gonna work a little bit differently. We're gonna work faster in some respects. We're gonna have more time for other things In other respects, maybe we don't do certain things anymore, like the manual laborious tasks.

Evan Troxel: Yeah. One of the things that I think about a lot with AI is that people just take it, take whatever it gives and spit you, spit that back out, and what you're

Rachelle Ray: Yes.

Evan Troxel: this is not mean automating stuff without oversight.

Rachelle Ray: No, please, please don't do that.

Evan Troxel: it's required, it's very required.

Rachelle Ray: Yes.

Evan Troxel: That's an oxy, it's an unnecessary, extra word there, but Yeah.

Rachelle Ray: But, but is it though, because I do see people just like take something sh GPT spits out as like, oh, that's hard. Truth.

Evan Troxel: Yeah,

Rachelle Ray: Let's run with it. I'm like, did you read this before you sent it to me? Because this is, this is totally wrong. I'm pretty sure this is not how this works, or this is not how we do things.

Evan Troxel: It's,

Rachelle Ray: Yeah,

Evan Troxel: I think. I think it, it's one of those things where. So how many times have you heard, you know, here's this fact about something happening in the world. Oh,

Rachelle Ray: right,

Evan Troxel: are you talking about? Well, I only read the headline, but

Rachelle Ray: right.

Evan Troxel: said.

Rachelle Ray: Exactly.

Evan Troxel: so you can't just skim this stuff.

Like you

Rachelle Ray: Yeah.

Evan Troxel: to get in, spend your time going through it. And so it saved you time building it, but you still have to spend

Rachelle Ray: Yes,

Evan Troxel: a good amount of time reviewing and going over it and maybe moving things around and rewriting things that you should, you, you would've had to do that anyway if you wrote it from scratch, but,

Rachelle Ray: exactly.

Evan Troxel: Yeah.

Rachelle Ray: Exactly.

Evan Troxel: another thing about this kind of internal versus external tool, you know, purpose built, a lot of people are using, like we talked about perplexity. You talk about chat, GPT, and these are tools that are available to everybody, right? And, and therefore they're not using the specific AEC jargon and they're not using.

And so you, you talked a little bit about RAG and um, I assume. With the things that you're talking about with, with the, the platform around OpenAsset and Shred.ai, is that because you service the AEC industry, that there's some, it, it understands vocabulary better, it understands these concepts in some other way because of the training that's gone into it.

Can you just speak to that a little bit just to give people some peace of mind or, or insight into why a tool being s specific for this kind of job might be better than a, you know, I don't know the right word. May not be generic, but you know what I mean. Like more avail, you know, widely available tool.

Rachelle Ray: Yeah, yeah. You're gonna get pretty good success with an off the shelf tool, right? Like you can dump a bunch of proposals into a chat GPT project, or a cloud project or whatever. You can query them and you'll have decent success. What you're going to see with a purpose built tool is. The difference between your marketing intern and like a graduate assistant.

So for example, I was testing this out, playing around with things. I put a bunch of proposals in chat and I was like, okay, tell me about, uh, the projects that we've worked on in K 12. And it returned a list and there were a couple of projects on the list that made me turn my head and go, where'd you get those?

Well, they were on someone's resume, but that was individual experience from a previous firm. And the ai, because it's off the shelf, it's not trained on this, didn't catch that, it didn't catch that this was prior experience, even though it's noted on the page and stuff. Um, it also doesn't really understand the difference between the prime and like sub or partner information.

So a purpose-built tool, like what we're doing is trained on that. We have given it definitions for this. We have done extensive labeling to literally show it like, here's, here's a proposal and here's. Information that would belong to the prime and information that would belong to a sub consultant, um, or a JV partner or something like that.

So that it's learning the way you would teach a, an intern or a graduate assistant, you know, um, what the differences are between these. So that's kind of like the major difference between off the shelf and built.

Evan Troxel: One of my concerns with, with this, with just AI in realm is that it, what it, what it could mean to affirm is just responding to more proposals faster.

Rachelle Ray: Yes.

Evan Troxel: I'm just curious what, what your position is on that, because I don't think, I mean, I mean, okay. Future work is necessary. More work is, I don't know, more maybe, maybe it's not more, but, but, uh, some amount of work needs to come in the door and that is happening through this process.

But do you see that, I mean, because we've seen like this proliferation of AI stuff on the internet, right? And it's

Rachelle Ray: Yes.

Evan Troxel: just, I would consider it junk, right? So there's a bunch of new junk, a bunch of new junk on the internet and a bunch of new stuff everywhere. It's like on Amazon, it's books, it's it's art in air quotes.

It's, it's all these things. And so does that, does that also apply to this? Or, or how do you see this actually being more useful, more effective?

Rachelle Ray: So number one, I'm seeing an increase in marketing of these tools towards that C-suite, right? Saying do more proposals faster with less,

Evan Troxel: hmm,

Rachelle Ray: really frustrating to me. Uh, having understood the tools and understanding the effort in marketing, uh, if I see another one of those campaigns, I will throw my computer out the window.

Please don't do that.

Evan Troxel: We'll put that gif on the screen right now. Like Yeah.

Rachelle Ray: Just, yeah, just I'm done with this because that's not how it works. Um, I am seeing an increase and it perhaps is procurement being impacted by the same AI push to do things faster and do more. I'm seeing more RFPs coming out. I'm seeing them come out with shorter deadlines, which is putting a lot more pressure on AEC marketers.

But I think what we need to be doing is using all of this AI power to better understand our market position, our ideal customer, where we're doing really well in the market, and where we're maybe falling flat. And start making more strategic decisions about the proposals that we're going after, the projects that we're pursuing, and maybe stop going after everything that we're probably okay qualified for, um, and really go after the things that we really want, that we're really well positioned for.

And take that time savings and one, invest it in more business development to really pre-position. For what you really do want. And two, really invest in writing strong proposals. Like I still see so many like volume shops where it's just like we just need to get as many proposals out the door as possible.

See how many of these contracts that we can get in and we're good. No, take that time savings, look at the ones you have the best chance at, and really invest in understanding that client and writing a really strong proposal. So that's where I stand. Um, I don't wanna see marketers doing more with less quote unquote, or doing it faster just 'cause AI is assisting.

Evan Troxel: I'm, I'm, I'm totally tracking with you because I think, you know, again, as an architect myself, there's a lot of jobs that come in the door that are just total morale killers

Rachelle Ray: Yeah.

Evan Troxel: they're not aligned with the type of work we want to do, the, the me the project and, you know, different levels of, of meaningfulness applied to different types of projects for sure.

But, and, and, and to your point, like firms do need a certain amount of work to come in the door to keep the machine running,

Rachelle Ray: Of course.

Evan Troxel: but doing the right kind of work is super important and, and it's often like not aligned with the incentive that people in business development have applied to their role, which is to

Rachelle Ray: Right,

Evan Troxel: work.

Rachelle Ray: right.

Evan Troxel: so if there's a way to actually, uh. Pull those things into alignment with these types of analyses so that the right strategy can be applied to the right kinds of work. It's more, it's better for the staff who's working on the projects too, right? So if to me that that's the twofer, or how, you know, you want to hit it on, on multiple levels on the spectrum to make sure that, that everything is being served in the right way.

Because firms that say yes to everything, like what is the value there, right? And I think people, people, one of my previous episodes, right, like people want to, it's their life. They're, they're not just investing in the company and they're not just showing up at a job, they're investing in their own life.

And you want to do the right kinds of projects. Like that's why somebody came to work there was to work with certain people or work on certain types of projects or get somewhere in their career trajectory that hopefully they could only get. That could happen there. Right. And so for all of that, for the leadership that it takes to kind of make that alignment and that strategy happen and, and really think about that proactively is super, super important.

Rachelle Ray: It is, it impacts everything. You know, one of the other things that marketing often has to do is market the company to bring in new talent. Right? And if you, you typically do that on, look at these amazing projects that we work on, or like,

Evan Troxel: that you're actually doing.

Rachelle Ray: yes. And so if you like hook people in, 'cause you've got this incredible like, mission critical portfolio and you bring in this architect and then you stick 'em on like bathroom renovations for the like local city hall or whatever they're going to, they're gonna leave real fast.

So yes, absolutely all of that.

Evan Troxel: Right. All right. Well, let's, let's talk about kind of the actual, uh, demand for AI and proposals. I think you're kind of making the case here that there, there is some demand, but can you give us kind of a, a broader view of what you are seeing for this kind of demand in the industry? Just, and I think it makes sense to kind of frame it under, what is our competition doing?

If we're not doing this, what are they doing, uh, that's out there. Because that to me is, is what's actually gonna inform what, what you're about to tell us. Mm-hmm.

Rachelle Ray: Yeah, so we did a survey at OpenAsset of, um, a couple hundred of our customers, and in that survey we found that 88% of them are currently using AI in some capacity. 55% of them said that they're using AI specifically for proposal writing. So that to me says that if your firm is not using ai, you're behind.

Because these firms are already playing with the technology. They're already getting a leg up on understanding the best places to leverage it within their firm, how to integrate it. They've kind of beat that. Technological learning curve of bringing in this new thing, right? And they're seeing the efficiencies from it.

So if you haven't even started, if your, if your staff isn't even allowed to, like open chat GPT or use, um, co-pilot or whatever you have available, then you're, they're gonna fall behind. They're probably gonna leave again, going back to staffing. They're probably gonna leave and go to a firm that allows them to use these tools to greater efficiency, to get the stuff that they don't wanna do out of the way so they can get to the good stuff that they do wanna do.

Right? AI answers your email, you get to design more. So that's what I would want to do.

Evan Troxel: Outsourcing. Yeah. the, the, I'm just trying to, to think about, there's firms out there who cannot. Let their staff use outside because of the types of contracts they

Rachelle Ray: Yes.

Evan Troxel: And so I'm just curious from your perspective, using a tool like the one that we're talking about here, because it's looking at an internal, you're building an internal platform.

Is that more, is that something that those firms who are bound by those types of contracts, uh, could use?

Rachelle Ray: So there are a lot of layers to that. Um, the most basic of which is making sure that your tech is soc two compliant. OpenAsset is SOC two compliant. So check good there. Um, there are federal regulations in the us Um, there are regulations for other government agencies as well. We need to be paying attention to those 'cause they're ever evolving.

But yes, having an AI sitting just on top of your stuff and not connected to the outside world is the first major step in making sure that you are allowed to use that information. You are allowed to use that ai. For what you work on. We are very aware of the fact that many, many, many firms work on projects that have confidential information or require clearances.

And so we are working to build in protocols within Shred.ai to help kind of segment that off, you know, put in permissions for who is allowed to see this. Um, sometimes RFPs even come in and they require information to be lock boxed away, right? So putting those permissions in to make sure that that is segmented and walled as it needs to be so that you know, we are serving our customers as they need to serve their clients.

Evan Troxel: Do, do you have, or do you know of any, any customers or you know, firms out there who are doing their own. LLM processing in-house locally versus sending this information out to the cloud for it to happen in a data center, which is, you know, I, I would imagine that's one of those layers that you were talking about, where it's

Rachelle Ray: Yeah.

Evan Troxel: it just can't go out and get computed somewhere else and then come back. It would have to happen on premises.

Rachelle Ray: Right. Um, I am seeing a number of firms trying to put together their own, like firm specific ai

Evan Troxel: Mm-hmm.

Rachelle Ray: super kudos to them because we have entire engineering teams to build Shred.ai and it's a really complex problem to solve.

Evan Troxel: Sure.

Rachelle Ray: So I think, I don't wanna like dismiss it, but I think that's a very difficult way to go because you need, this technology moves so fast and if it updates it, if, if it evolves and you don't update and you don't evolve with it quick enough, then you open yourself to vulnerabilities.

So I would, I would be a little bit worried about risk and security there.

Evan Troxel: Yeah. Yeah. It's, it's definitely a huge investment

Rachelle Ray: Yeah.

Evan Troxel: do that because like, and, and think about who you're competing against,

Rachelle Ray: Right.

Evan Troxel: because you top talent in the whole world when it comes to that. And architecture firms don't specialize in this, and it's hard enough to write your own and maintain your own software, right?

Rachelle Ray: Right.

Evan Troxel: firms aren't set up to do that. And so, um, I'm just curious to hear if, I was curious to hear if there was any, any firms that you know of that are attempting to do that. And, and not to name them, but just if that's happening, because I think, you know, that's a. Possible solution, but it's also a very difficult road to go down.

Rachelle Ray: For sure, for sure. I can think of one specifically, um, where I was talking to a team at a conference earlier this year and they're like, yeah, our IT team put together an AI that is for marketing, but not all of us are allowed to access it. And those of us that can aren't really happy with what it's doing, but like it doesn't understand what marketing does.

So like. They thought they built a really cool LLM that's just looking at their data. But again, coming with that purpose built, does it understand partner information? Does it understand the nuances? Um, it wasn't quite working the way they wanted it to, but from like a firm general perspective, everyone was like, oh, this is kind of cool.

We have our own thing. So again, you know, mixed success, mixed bag. I'm curious to see how that's going to continue to evolve, but I think it will be, like you said, think about your competition. Think about who has the resources to do that in-house versus what resources you have available to do in-house.

Evan Troxel: I think it's happening there, but, but I think that's gonna be severely limited to the largest of, you know, the large

Rachelle Ray: Yes,

Evan Troxel: Who have the resources to get to dedicate to that and the, the capacity to have people working on that. I, I wanted to shift gears and kind of go back to something you talked about freeing up kind of time

Rachelle Ray: I.

Evan Troxel: strategy. We've, we've talked about on the podcast before, like, time to free up for creativity by taking away kind of this more mundane stuff.

Rachelle Ray: Right.

Evan Troxel: I'm just wondering if are people actually doing that? So, so maybe let's start off with what are you seeing as like the, the time savings? If you could give an example of like what traditional marketing business development proposal response, like. Let's say three years ago, what did what, how much time did that kind of average take versus what you're seeing now with these new tools?

Rachelle Ray: Sure. The easiest one to identify is like reading the RFP, right? So historically, for me to read through the RFP and then to create a proposal management plan where I pull out all of the key information, what do I need to know about the client, the submission, the formatting, the response criteria, all of that, that could take me anywhere from like two hours to an entire day or more, depending on how big the RFP itself is, how big the pursuit is.

I took a week once on a multi-billion dollar pursuit just to figure out what the heck they were asking for and how many different parts and pieces there were. With ai, that literally happens in, you know, five to 30 minutes, right? Going back to AI land and your, um, level setting on how long it takes to do things, it can read through these documents and extract that information very, very quickly.

And yes, you as the human should still be looking at what it has provided you double checking it against, but it can kind of cut through that contractual language. It can cut through all of the extra fluffy stuff that's in, uh, your RFP and get you just to the meat of it very, very quickly. So that's a huge time savings.

You know, eight hours to 30 minutes, I will take that two hours to 30 minutes. I will take that.

Evan Troxel: Hmm.

Rachelle Ray: Um, the other biggest thing that we saw in our survey of customers was that almost 40% cited searching for past information, searching for that content, looking for project information, for their proposals as their biggest, uh, time suck.

Evan Troxel: Yeah.

Rachelle Ray: I feel that, um, time to first draft is always the longest for me. It can take. It can take 10 hours to two weeks, depending again on how big this thing is. I'm gonna average it at, you know, we'll say 12 hours for me to get to my first draft. 'cause I have to pull in all of the pieces and kind of start getting them into place.

And that's not even massaging them into the story or starting to like weave things together. That's literally just me pulling in boilerplate, sticking in resumes, grabbing old project sheets and getting them in place in my outline. Ai being able to crawl your proposal library can do that so much faster, right?

You can probably generate a draft in a couple of hours working alongside with your ai. Like, Hey, here's this first question. Go crawl my library. Have I answered this before? Surface? Drop it in. Next question. Let's go. So that's happening in, you know, seconds, minutes as opposed to me clicking through folders.

Like you said, they're very opaque, going through old proposals and stuff like that. So that's, that's from seeing the most time savings happening.

Evan Troxel: Yeah. Like those, those folder numbers, right? The job number is, that is computer language. It's not human

Rachelle Ray: No.

Evan Troxel: Right. That alone is, is, is an amazing thing because Yeah. And I, and I could see that being, it's not just super valuable because, oh, it, it cuts the time down. But because to me, like you actually get to ask it questions and, and you get to kind of talk to

Rachelle Ray: Yes.

Evan Troxel: proposal in that way, I think that's, that's pretty interesting. This has never happened, right? There's never been conflicting information in an RFP,

Rachelle Ray: Oh, never, ever, ever. The client always double checks.

Evan Troxel: to find that information is difficult. Right. And, and it usually happens like way down the line because, uh. A lot, especially during design build projects, right? Like the, the kind of RFP response has a new design in it.

And so

Rachelle Ray: Right.

Evan Troxel: team is kind of working in parallel with the marketing department to develop response. And

Rachelle Ray: Right,

Evan Troxel: there's prototypes, there's like early design, there's all these things, and then you get down the road and it's like, oh, did anybody read this? And it, it completely says something.

It says something completely different than, than what This section over here. So are you seeing this tool even being useful to identify conflicts that, that are in the, that never happen in

Rachelle Ray: for sure. Um, Shred.ai specifically is created, developed so that it can flag those and service them to you. Like, hey, you know, super easy. One client said this is due October 17th and October 21st. I don't know which one is real. You might wanna ask surfaces that, that conflict. Um, our little chat bot that follows you around the platform can interact with that document specifically, and you can say, Hey, like, what risks do you see in here?

You can do that with chat GPT as well, again without the AEC understanding. But it will look through, it'll read that scope for you and it can do the same thing and say, Hey, in this paragraph the client was really talking about sustainability. They cited a lead gold, uh, goal for this project. And then later on they said that lead is not a, a component of this project.

So hey, which one is it? Which affects your strategy is marketing, right? Like I'm probably going through

Evan Troxel: turns out they're copying and pasting from other projects as well. Yeah.

Rachelle Ray: Exactly. Exactly.

Evan Troxel: that, that kind of stuff does show

Rachelle Ray: Super helpful.

Evan Troxel: seems like a pretty helpful use

Rachelle Ray: Very.

Evan Troxel: Anything else as far as like freeing up, um, the ability to be creative, strategic alignment with project types to the firm?

Anything else like that that, that you're really seeing winning, like firms winning with, know, whatever they're telling you.

Rachelle Ray: Yeah. I'm gonna loop it back to a conversation we were having at the very beginning with talking to our, our SMEs, like getting that collaboration component because that's where I reinvest the time back, right? If I don't have to spend 12 hours hunting for information on the server and I have my outline, I can now spend more time sitting with my project manager, the designer, whoever it is, and having that casual conversation, because I know y'all don't like the blank page.

I know you're not writers. That's okay. That's my job. But if I can just get the architects to talk to me, the engineers, whoever it is, tell me about like. I don't even ask about challenges or solutions or anything. I say, what pissed you off about this project? Like you're just talking to me. Yeah,

Evan Troxel: Yeah.

Rachelle Ray: you're just talking to me.

Vent, complain, and then I marketing speak. That becomes my challenges, right? Or like, what are you super proud of? Did you have a moment on this project that you really loved? Or like if you were taking your family to go see it, what would you show them first? Having the time to really practice those interview skills, really understand those internal stakeholders and how to talk to them to get those insights.

That's where the magic is gonna happen. So AI giving you the time back to do that and then giving you the time to take those insights and really work with them and be creative and do cool things within your proposal. That for sure is, is the whole point of all of this, right? It's not to do more faster, it's to do better,

Evan Troxel: deeper, deeper

Rachelle Ray: deeper.

Exactly.

Evan Troxel: Can we tie this back to kind of just tools real quick bolts,

Rachelle Ray: Of course.

Evan Troxel: and bolts. That's the right words. Um, so we talked, you, you said early on PageMaker to InDesign. Right. And, and I'm just curious how a tool like this. Fits in with that. So are you doing all of the work in this and it's actually generating the proposal?

Are you these documents into something, a page layout program like InDesign now? Or how are, how are firms actually pulling this together these days? Mm-hmm.

Rachelle Ray: Um, so I don't know very many tools that integrate with InDesign. OpenAsset does. Um, you can spit out resumes and project sheets right now out of box with our OpenAsset Dam into your branded templates, which is great. Shred.ai will export to InDesign. But you're going to start your drafting within the system so that it can interact with your proposal library.

You were talking about, like interacting with it, querying it, you know, massaging those narratives and stuff. You're gonna do all of that in the system and then export it and then you have time to like tweak your layout and such. Um, but you need it to stay connected to your insights for as long as possible to really build your outline.

But yes, yeah, we want it to meet marketers where they are and integrate with the technology that you're already using.

Evan Troxel: And I think that that gate there is, is a, is an extra safety net, right?

Rachelle Ray: Yeah.

Evan Troxel: I have, like, I use Notion as my tool of my database of just my life. Everything's in there, my podcast information. And one thing that actually annoys me about it is because there's certain AI functions that can just be plugged into every page.

So, so for example, like if I have, uh, show notes for an episode, like this one in there I put something new in that page, the AI is gonna go back through and kind of recalculate and readjust. And I don't, I don't need it to do that all the time. And so I

Rachelle Ray: Right.

Evan Troxel: a break between these things at some point so that I know what's in there is going to be what's in there.

I don't want it to be kind of processing that information all the time and, and rewording it or coming. No, absolutely. Especially with something like this, I wouldn't want it to be messing with that. I wouldn't want a live link. Back to, you know, that LLM. So because that, that to me presents a, a risk,

Rachelle Ray: Mm-hmm.

Evan Troxel: the language. Oh, update the information because somebody pasted in something maybe they didn't even mean to, right. But then

Rachelle Ray: Right.

Evan Troxel: taking that into consideration and maybe updating a summary or something like that. I don't want it to do that. I, I

Rachelle Ray: Right.

Evan Troxel: a break in there. So I think that this is a, a, a nice gate to have in the system.

So it's like, okay, I'm gonna do this work over here in the sandbox, and I'm gonna put it over here into the thing that the client's actually gonna see.

Rachelle Ray: Yeah, exactly. And that's just another human in the loop, right.

Evan Troxel: Right.

Rachelle Ray: well. It's another place for the marketer to intervene and say, all right, should, I'm cutting it off. We're moving into InDesign. I will take it from here.

Evan Troxel: Yeah. Yeah. Cool. okay. So I, I guess to wrap it up. One of the, my big takeaways from this is, and this is something that's I think apply applicable to a lot of AI stuff, which is, you know, there was a huge focus on prompt engineering, right? And

Rachelle Ray: Right,

Evan Troxel: what's the right way to talk to an ai? And I, that's changed pretty significantly in the last two years.

And now it's much more about context,

Rachelle Ray: right.

Evan Troxel: And, and I think that's kind of probably the new

Rachelle Ray: I.

Evan Troxel: of prompt engineering is context engineering. Because I think a lot of people use AI and they ask it questions and they're like, well, the answers sucked, so I'm not gonna use that anymore. And it's like, well, that just because you asked it a question doesn't mean it understood the context.

And so the more context you can give it, the better results you're gonna get for a conversation about something. Not just, I'm not just gonna put in a search and expect the answer. I'm going to work with it to look at different sources, consider things in different ways to ask me questions. Um, those are all great ways to kind of interact with the. an LLM. So, um, when it comes to surfacing content with the right context and kind of working with people to prompt that information and keeping humans in the loop, uh, it sounds to me like that's what you're doing with Shred.ai. Is that, am I in the right place here?

Rachelle Ray: Absolutely. We're trying to keep it as intuitive as possible for any user. I want a marketer to be able to go in there and query the library. I want an architect to be able to go in there and query the library and both be satisfied with the results that they get. So we are, you know, doing our magic in the background, um, to build in those guardrails, to train the Shred.ai, to understand context, teaching it, you know, what proposals look like, what RFPs look like, but also giving it that power to go back to the user and either redirect them or prompt them back and say, Hey, you know, thank you for giving me this input.

I can return a better output if you do X, Y, and Z. So kind of building those in so that the user doesn't have to troubleshoot and figure out, you know, like back in those chat PT days, I'm just an LLM, I don't know what to do. And you're sitting there like, okay, well maybe if I ask it like this or this.

Evan Troxel: right.

Rachelle Ray: We have put that guardrail in place.

Like, okay, this is a new type of tool. It's a new, you know, thing for you. It's a new thing for us. Um, here is how it's going to tell you how to use it. Right? That's like the ultimate technological thing is the tech teaching you how to use the tech, but intuitively it's seamlessly. So that's what we're doing with Shred.ai.

Evan Troxel: Cool. Well, thank you for taking the time to have this conversation. I will have links to everything that we talked about in the show notes for this episode. So if anybody out there wants to learn more about this, there'll be a couple different landing pages that you can go visit to learn more. Is there anything else that we didn't cover that you wanted to make sure we, we hit on before we stop?

Rachelle Ray: No, this has been fantastic, Evan. Thank you so much for having me on to talk about Shred.ai and AI in general and the plight of the AEC marketer. Thanks for listening.

Evan Troxel: Well, thank you for your techno optimism. I think there is a lot of doom and gloom when it comes to this, and I, I think it's reasonable to, to have that, you know, but, but also to hear, you know, the potential behind it and where you see this going and, and I think it's great to just let people know what's actually happening out there.

And for me, like, you know, we talked a little bit about the AEC industry being behind the, the technology industry for sure. Right? The bleeding edge,

Rachelle Ray: Right.

Evan Troxel: fine. Um, in this case, I think that's fine because I think it's our job to find our authenticity through this and not

Rachelle Ray: Yes.

Evan Troxel: And, and doing more with less just for the sake of doing more with less.

So, um, I think that, that, that trepidation is warranted. And also there's, there's a thread to this conversation that is an enabler for people to be really strategic and aligned with the kind of work that they're going after. And tools like this can actually help surface that information early. So in that go, no go process, right?

You're, you're just saying yes to the ones that, that really matter and they can, that your firm are really well suited to, to go after. So

Rachelle Ray: For sure.

Evan Troxel: the conversation. Thank you.

Rachelle Ray: Thanks Evan.

​