Full Transcript
[Music] Welcome to the Solspace Podcast. Thanks for listening.
Mitchell:
Everybody, welcome back to the Solspace podcast. This is Mitchell Kimbrough, your host, founder of Solspace. Rejoining the podcast is Matt Everson from Astuteo.
Astuteo is a web development agency very much like Solspace. We do a lot of the same types of work and Matt and I talk on a regular basis sharing notes. We even have a little bit of overlap in the market segments that we serve, but as is the case with most web development agency owners, we get a lot more out of being friends than enemies.
So I count Matt as a good friend and colleague and advisor to me. So let me take a minute to thank you for that, Matt. You've been really great help to me over the years.
Matt:
You're welcome. Same to you.
Mitchell:
You and I just returned from a conference in Austin, Texas called the Industrial Marketing Summit.
It was my third year there, your second year there.
Matt:
Yep.
Mitchell:
But you and I both have clients in the manufacturing space. Almost everybody at that conference is in a marketing seat, marketing director, VP of marketing, marketing lead of some kind. And almost uniformly across the board, these people that we're talking to at companies like that, they're solo, solo practitioner.
Maybe the manufacturing company has 100 employees, 150 employees. There's one marketing person and they're juggling all the marketing tasks and channels and they're doing price lists and they're updating the website and they're updating LinkedIn. They're doing so many things all at once and they all have the same look on their face.
They come to this conference and they get to meet one another and talk to one another and just share war stories. And I really enjoy watching them do it because the sense of camaraderie that they get there is, they just have a lot of support from each other. And you and I help people like that.
And we get to have conversations with them to find out what hurts. And sometimes we can, we can step in and give some advice. And lately the advice I'm giving to a lot of my marketing director clients is jump head first into AI.
And anyone who knows you or watches you on LinkedIn knows that you've been talking a lot about your use of AI. And that's going to be part of what we touch on here. But the specific topic I'm hoping to get into today is what I think you and I have experienced a number of times where in a business, in an organization, when they want to do something digital, maybe the organization wants to develop a new product, a new version of how they make their services available to their customers, to their clients.
If they want to take it online, maybe a subscription service of some kind, maybe make it available on the web in a way that was not something they had done before. And it's not strictly speaking a marketing activity or a marketing product. They nevertheless route the request through the marketing team, through the marketing director.
And that marketing director comes to us, the people who build stuff for them, to say, hey, could this be done? And the frequency with which I get requests from the marketing directors, marketing clients that we serve, who are asking us to build stuff that's not strictly speaking part of their marketing initiatives, it's really surprising. But then it's not surprising, because who else would that organization turn to other than the person in the building who has the most seasoning and experience with digital tools?
So does that ring a bell at all? Have you encountered this in your work?
Matt:
Yes, and we're a little bit different. I mean, you are very similar, like you said, Solspace, but we're different in that unlike a typical web design and development company, we are mostly in manufacturing and industrial.
And so we end up building things like customer care portals, integrating with ERP systems, integrating with Salesforce. So that stuff is generally crossing over into sales and operations anyways. But to your point, I think it's not even that the sales or operations are coming up with these ideas alone and then coming to marketing.
These marketing directors are sitting in those meetings, someone brings up an idea, and it's just that is the position that is connected to the most people who can execute that idea, which is typically their web design, development, digital team.
Mitchell:
I had a meeting just this morning with a client who's been on retainer with us for a number of months. We came in to basically rescue a website that had been poorly built. It had been hacked.
It was performing terribly. It was difficult to update. Fortunately, they were on a good tech stack.
That's how they found us. We cleaned that mess up, and my client, VP of marketing client, came and said, all right, well, I guess stuff is fixed. Maybe we don't need you guys anymore.
But before I say goodbye, I thought I would give you an opportunity to tell me what else Solspace is capable of doing. And it probably took five minutes of me saying, one of the things we're really jumping into is using AI as our tool of choice for doing our job. And here are some examples of things we've recently done that we did 20x faster than we could have done four or five months ago.
And that prompted a lot of conversation. And the very same thing happened there. This VP of marketing said, oh, that gives me an idea for a new revenue stream for the whole company that the CEO is going to hear from me, the marketing lead, that came out of a digital conversation.
And really, the primary thing that was happening was, here are some additional tools that you didn't know were available. Here are some additional materials that you can build with that they're new technology, and you're the first to hear about them because you're the closest to digital. So with respect to AI, Matt, as you're having conversations with these clients, what sorts of things are you seeing possible?
What kinds of capabilities are you bringing to these conversations?
Matt:
Yeah, good question. I think right now we're in an especially strange time, like Claude Opus 4.6, the sort of beginning of February. Our conversations right now are sort of like me on a laptop meeting with the marketing director, pulling up Claude Code and just kind of barfing all over them.
What's now possible. And spinning up as many examples as I can, including like, let's scaffold a Craft CMS site, you know, with Alpine and Tailwind and, you know, a Vite build process, everything end to end while we talk. You know, two days of work sort of thing, 12 hours of work in the next 10 minutes.
You know, and with that, showing them some estimating tools and whatever it is, like a handful of different skills that are possible, which we're building in a manic frenzy now every single day. And I think every web developer's in the same boat. But to your point about them realizing what's possible, when I have these kind of semi-crazy conversations these days, the light bulbs just start going off like crazy with the marketing director.
They start saying, well, you know what, I just had this idea and I wonder about this department and I wonder about this. And, you know, I'm just like, yes, you know, so it's like, it's trying to get them just a little bit up to speed with where we are. And once they see that their ideas are basically infinite, the same as ours are infinite.
So the way I see that progressing over the next, you know, 30 to 60 to 90 days is those web developer marketing conversations turning into larger conversations with the organization simply for the sake of more of those light bulbs going off in more departments' heads. Because we've talked about this. This is the way everything is going to go over the next few years.
And it's going to be an incredibly rapid change in knowledge, work and operations across every business.
Mitchell:
The phenomenon you're describing has been a great relief to me. A few months ago, and by few, I mean no more than four months ago, had you sat me down and we were talking about what AI is going to do to a digital agency like mine or yours, I would have shrugged my shoulders and flipped a coin and said, could destroy me and I'm going to go flip burgers or it could make things explode.
I don't really know. I had the same feeling about the pandemic. When the pandemic hit and people are like, hey, what's going to happen?
I don't know. Let me flip a coin. We could be nuked completely.
There could be no business or we could go completely crazy. We went completely crazy.
Matt:
Do you remember that is exactly what happened. It was like this very dead period before this huge tsunami of work when everyone realized digital was now their only channel to reach people.
Mitchell:
Yeah, that's right. So digital was the only channel. And I don't want to force a comparison between digital, the pandemic and AI digital.
But there's something similar happening in so far as tsunamis go. I appreciate you saying tsunami because that's exactly what I'm looking at right now. So you were saying your conversations are all taking the character of you can do that that fast.
Can you do this?
Matt:
Yeah.
Mitchell:
Yeah, we could do that. Could you do this? Yeah, we do that.
Could you do this other thing, too? Can you get can we land on the moon? Yeah, probably.
Yeah, exactly.
Matt:
Right.
Mitchell:
So but you you can't say that unless you've had the courage as a developer to sit down with Claude Code and say, OK, go ahead and blow my mind. Tell me how the future is going to be. Do I have a place in it?
I've been very relieved to see that everybody that I just speak absolutely bluntly, transparently honest to about how fast we can go now, how cheap we can build now. I'm just like, look, I don't know what's going to happen after I tell you the truth, but I'm just going to tell you the truth. The answer is always what you said.
The answer is always, wow, I have 15 ideas that I was too scared to tell you about. I didn't even have the ideas until you started telling me some of these capabilities. And now I have brand new ones.
Right? So there's just going to be this unlock of massive productivity. Some stuff is going to be built that could have never been built before.
Matt:
Right.
Mitchell:
So that's really the pattern that I'm seeing now. How do you think it's changing?
We stray a little bit from the topic, but how do you think it might change staffing as far as how these marketing directors are thinking about who they hire, what they spend, who does what, where the stuff lives? Any thoughts on that?
Matt:
Yes, I have a lot of thoughts. I don't know if they're on that exact question, but like what I was telling people at IMS is you have to be the person that is trying to automate away your own job. Because you are the person who, you're the only person who knows how to do your job.
Typically, we all are, right? Like other people can do the job, but that information is trapped in your head. And it's much harder for you to go and try to automate HR if you're not in HR, but you take your own position and you begin using these tools or using a partner, you know, like someone who knows these tools to start automating the smallest thing.
You know, and start connecting these parts until your job is effectively gone because your job's not going to disappear. You're actually going to move into the position of being the AI forward leading person at your company or one of those on that team. As far as staffing, like as far as what happens to jobs, all I know is things are going to be incredibly automated at a really fast scale because you and I know we're working in this stuff.
It is not hard to build this stuff out. There's a lot of structural things that developers and software engineers understand intuitively or through experience that helps them build these tools efficiently, but it is not hard to build this stuff right now. And so if I had to guess, and we have no idea, there's no way of knowing, but I think eventually things will look fairly similar.
The people who are just completely anti-AI are not going to have jobs because they'll get automated out of jobs and they won't have a new position to move into. But I think, like, I can imagine a scenario where you have a small internal team that is actually web development, AI oriented, that is building out your processes. But I think that will start with companies like ours coming in and consulting on and having these meetings and showing these teams what is possible and then helping lead them through that initial execution.
Like, how do we get this started? We will help show you how to get this started. And maybe three years from now, they're like, you know what, we need a three person AI/IT/marketing mix team or whatever it is.
But it's this team that manages the automation of the business.
Mitchell:
What you're describing of being AI forward, AI-First, being proactive about how you automate away your job because you know your job the best. This implies an underlying faith in there being something for you after. It's, you know, it's going to be a Mitchell, you can pick up Claude Code and you can build the thing really fast and you can turn it over to your client on Monday after talking to them on Friday.
It's all right. Everything's going to be fine. They're going to love it.
They're not going to freak out that it now all of a sudden they can cut their budget by 300 percent. There's something on the other side. So this implies a faith.
Now, this is something I see in my marketing lead clients on a regular basis. I think there's some kind of a natural selection at work with people in the digital marketing space. The ones who were scared to adopt a new marketing tactic, scared to get into a new channel, scared to jump into TikTok and try it out, just experiment, throw it in the garbage if it doesn't work, try something new.
The ones who were not comfortable doing that, I don't think they flourished. So there's something about the wiring of these people. Not only are they connected to digital technology because of their job, but they have to be comfortable trying experiments, failing, trying something else over and over again in the digital space.
And that's part of what this AI adoption is about. Let me just try this thing. I wonder if this could be built.
Oh, that sucked. Let me throw it away. Let me give a better prompt to Claude.
Let me give Claude a smaller piece of that to do first, and then it can do the next thing. Right? So it's an experimentation mentality, and it's sort of an abundance mindset that these people tend to flourish with.
Now, is that culturally in these organizations? Are people coming to that marketing lead because they're the ones who have the creative problem solving capability as well? Like, is that a component of this?
Matt:
Yes, of course. I think that is one of the more creative thinking roles in the business. But I was thinking about this earlier.
The reason marketing is going to be the catalyst or like probably the primary catalyst for this whole operational change is because if you think about the role of marketing, especially the marketing director or, you know, the people that are directly attached to digital is in both cases with digital and with marketing in general. To do the job well, you have to try things and you have to fail. You know, you're always, let's try this new application and spin it up and see if it works.
And if it doesn't work, toss it in the garbage. Let's try a new one. Let's try a new campaign.
Does it work? Nope. You know, so you're always testing new things, trying new things.
It's a creative role by nature anyways. HR can't go test something or try something. Finance can't install some new software and abandon it, you know, after a two-month trial, after they've moved all their systems into it.
So marketing is just naturally the creative sort of R&D, what should we do next as a company role in the business. I think it's going to be a much, much bigger role in this transition than it has been in the past. But I think it is naturally that kind of role.
Mitchell:
What are some of the things you were, you were giving me the example of being on, you know, your recent experience of calls with your clients. It's just been furiously coding in front of them, the building stuff on the fly, just trying to ignite creativity and just seeing what's possible. What kinds of things did you build?
Like, are there patterns that are recurring?
Matt:
Well, so, you know, this stuff is so new. I just feel like we have to put it out there and show people what is going to happen, because I've written a couple of articles where it's just, I think this is unavoidable and is going to sweep through all of American businesses in the world. But, but definitely America being so knowledge, knowledge, work oriented.
If you sit in front of a computer, your job is going to get extremely, if not wholly automated. But I don't know. Like, so we build all these skills.
We're mostly automating our own agency. So things like, you know, I have a voice DNA skill that will draft something like a LinkedIn post in my own voice. It'll draft a draft version and a final, I'll edit the final to be my actual post, save them.
And it automatically then looks at the differences between the two and updates that voice DNA file so that in theory, I'm always moving closer and closer and closer to my exact voice, which I don't think will ever be possible. I'll always be making edits, but it's gotten substantially better. Like I said, things like just spinning up a multi-day sort of scaffolding project setup.
It does that. Other things like I will, I'll dump in the meeting notes, you know, like the transcribed AI meeting notes from a meeting and it will, it's like a meeting extraction skill where it will pull out all GitHub related issues. You know, so it knows our clients.
This is all sort of on an MCP server we've developed. It'll pull out GitHub issues and put those on a list and send them to GitHub. It will pull out follow-ups by individual, like who needs to do what it'll pull out decisions that were made in the meeting.
And some of this, I don't really even have a place for it to go yet, but it'll do that. Other things like we've built a heuristic based project estimation skill that I just spent a half day sort of explaining how we hip shot project estimates and it's dead on perfectly accurate. I would, I would more or less be comfortable putting it, putting it on our website and just letting estimates get auto approved.
So again, I will drop in meeting notes. I will run the SOW generator. It's especially accurate if it already knows the client and past projects and related GitHub repositories.
But even if it doesn't, you know, 13, 18 seconds later, it outputs a formatted SOW with a scope and a price on it because we have our entire services registry as another file. And, you know, it's yes, we do web design, web development, like a few buckets of services that services registry has like 37 or 38. It keeps adding as it realizes we do certain little buckets of different things for clients.
It adds to that services registry and sort of builds its understanding of what we're capable of, which it can then use in scoping projects. So anyways, I could literally make this like a four hour podcast if we keep going on skills like that. But that's, that's a lot of it.
So I just, I do the same thing. I pull out my laptop and like I said, I just start puking all over the client, spinning up, you know, new tabs and showing it what's possible with all these skills. My, the last meeting I had, he's like, this is really great.
And I'm going to put you in front of some more people on our team, but, but you need to refine, refine this presentation a little bit.
Mitchell:
So those, you were touching on a new question I have that's related to this topic and that is, all right. So you, you're a marketing lead at a company of some kind, maybe manufacturing, maybe some kind of service provider, maybe a subscription service, a SaaS. So there's all sorts of variations on this theme.
You know what the new building materials are just because of who you are. You know that the new building materials in the digital space are the following. And because of that, you using that, that tool set and turning AI loose on it, you can accomplish crazy stuff at high speed, reliably, good quality products, launch them quickly, tear them down.
If you don't like them, rebuild them. What are those underlying tools? What are the underlying platforms?
What is AI really good at building in? And what are these marketing directors and marketing teams need to be thinking about? Just take the example of, you know, we had some conversations this past week in Austin where I was watching marketing directors have that struggle of, I know I need to migrate off this platform.
We're on WordPress. It keeps falling apart. We got a thousand plugins.
I need to be on something else. What should I migrate to? So what kinds of tools do you think you could list now that these marketing teams need to be thinking about and having it as part of their, their ongoing vocabulary?
Matt:
I don't know that I would list like specific content management tools or things like that at this point. I think once you get into this, you do realize that some of the newer content management systems that are, you know, they might be headless systems or actually, you know, sort of the bulk of them created after 2022 or whenever AI came about. They are just structurally set up better to manage things through the file system rather than the database.
If AI can scaffold all the fields and the sections and the entry types and all the, and the global variables and all of this through the file system, it's radically more efficient than trying to do a database, you know, build a database migration file with AI and then run that through the database to generate the artifacts that come out the back. So that part is definitely, that's just something we are trying to watch and figure out ourselves.
We're not changing anything right now, but we're definitely aware of it. I think this isn't exactly what you asked, but I think the thing that the latest AI models are just incredibly good at is process. And particularly like when you develop a standard operating procedure, five years ago, lots of companies have tried to do this, especially in manufacturing and skilled trades, where a guy that's been working there 35 years retires with everything in his head.
And it just walks out of the building. Companies have been trying to figure out how to document some of this tribal and institutional knowledge. And they've been doing it with like word documents and there are tools like process street and different ways to capture that information.
They've been trying, and it's just a very hard thing because you end up using a 44 step process for the third time, you know, in three months and you realize step 12, 18 and 37 are no longer relevant. Something's changed. You now have to go back and you have to update that standard operating procedure.
Well, you don't have to do that with Claude and these latest tools. You can build out this absolutely massive knowledge repository. And if you come along tomorrow and say, Hey, you know what, what I told you yesterday for step 37 is now we're going to use this tool instead.
Claude will on its own without any requests say, okay, we just updated that. I'm going to go through this whole repository and okay, here are 14 instances where other processes, skills, conventions, what have you reference that particular step. And I'm going to update all of those too.
So it process-wise, it keeps this sort of organizational brain completely up to date, which has never been possible at this scale before. And by scale, I mean, both such a huge scale and also such a small scale. Like any business where we are a very small business, we have all this capability.
Mitchell:
Yeah, that's right. That's one of the things that I think these marketing leads need to be thinking about is those who have really massive agencies on retainer or at their disposal and have had for some time because of the capacity that those agencies bring in the mix of expertise that's under one roof. That's going to get nuked by AI.
Someone who knows how to use Claude well and as versed in a number of tools that Claude is comfortable working with, we can scale. I mean, I'm seeing some posts about the optimal agency size is drastically shrinking for this reason. And that's scary for those who run larger agencies with a lot of headcount.
I mean, they have sales teams that'll keep the lights on. But for people really solving problems and marketing leads who really want an efficient, cost effective solution to a set of issues and want creativity, the AI-First agency is going to have a massive advantage. The question of tools and materials, you know, I've, I've pointed Claude Code at, you know, legacy installs of WordPress, and it has been capable of building from scratch a new version of that very same website on something like next JS, maybe a headless Craft or a strappy or something.
Building from scratch replacement for that website, faster and better and easier than if I asked it to work within that code base. So there's, there's some winners and losers out there. Like this morning, I had a conversation with VP marketing, and I was making the case for putting an AI like a custom coded AI agent in front of their customers as a new web app.
And that agent would have RAG capability. So retrieval-augmented generation, the winners and losers in this are the Algolia's and the Meili searches as a simple example of one of many things in the toolkit, you put your product catalog or your, you know, whatever kind of database of records that you have into a system like that, that's headless and API accessible. That's super, super fast.
And it has, has been built to make a faceted type of an API search capability available. And AI loves to work with that other forms of databases, not so much. So there's, there's some winners and losers here.
And it's interesting to be thinking about and I like when my marketing leads to the people I work for. I like when they understand the tools at least a cursory level. So that's, those are, those are some of the thoughts I'm having about AI.
AI is super great at some stuff. But if you ask it to do some other stuff, not so good.
Matt:
Yeah, I was talking to someone at IMS and that had a, you know, fairly substantial startup and he had hired an agency to build a site. They put it on Webflow, a nice looking site. He was not that familiar with Webflow.
He's also not a developer. He went in there and, you know, he would try to change a card layout or something and he would blow up the whole site. He eventually got frustrated enough and is intelligent, definitely a super intelligent guy, but he just started working with these latest tools, ended up on Next.js and I think using Sanity CMS headless and totally admitted he really has no idea what he's doing. It showed me the size of his site, both content-wise and quality-wise. And it's incredible.
And so I had kind of arrived at a similar place coming from what we use and trying to figure out what the more AI-First alternatives are going to be. And it was just interesting to see someone else sort of arrive in the same space. But I wanted to talk tech and he's like, Oh, I have no idea really how this is set up.
He's not a software engineer, but you know, it's, it's the same thing you're saying where this stuff is just totally possible now on the right platforms.
Mitchell:
I had the experience of a friend and, and colleague come to me last weekend. His son has a startup idea. They wrote a business plan.
They sat down with Claude Code and they uploaded the business plan and said, Hey, build the app. And what they got was unusable. And then they came to me and said, maybe we should ask you to do what we just did, because you know how to build stuff on the web.
Like you would probably be better at prompting Claude Code to build a thing, a big complicated thing, because you know what to ask and you know what to tell it not to do. And they're exactly right. I mean, that's what I'm seeing.
We do have experience with the tools, we understand what's going to happen when this thing is under load under a traffic surge, get a bot attack, like all these different, you know, the future-proofing of some sort of a content model, all those things we have experience with and we can help Claude make good decisions and we can break the thing down into smaller Claude sized pieces to solve. So there's still a reason to hire us. These marketing leads don't necessarily just dump us and go to Claude Code and say, build me something good.
When things get complicated, it's good to have someone with some experience who can guide this very productive tool.
Matt:
Yeah, absolutely. I think so.
Mitchell:
Well, maybe that's a good place to wrap up. I think we need to keep this conversation going because if we did this recording next month, things will have moved. You know, it's at the speed of lightning.
Things are moving so fast. That's one thing that's a relief to me is nobody's an AI expert. Anybody who claims to be an AI expert go away, leave me alone, because it's moving so fast.
Nobody, nobody is.
Matt:
Oh, I know. And the ones who said they were a year ago and went all in on custom GPTs or something else. I knew at the time it made absolutely no sense to build or invest in anything like that.
At least not try to put a foundation down. I do think that this operational level knowledge and process automation is just a fundamental, huge change in the way business is going to run. So it's definitely something to build a foundation in.
Mitchell:
Yeah, I agree. Well, Matt, thanks for making time. Always good to talk to you and get some good insight and send me off in good productive directions.
So we'll keep our conversation going. Thank you again. Where do people find you online?
You know, what's the domain name? Give a plug to your studio.
Matt:
Oh, sure. The company is Astuteo, spelled like the word astute with an O on the end. So it's just Astuteo.com.
And I am on LinkedIn, Matt Everson, and my email is just matt@astuteo.com.
Mitchell:
Good. Excellent. Thank you, Matt.
Thanks, Mitchell.
[Music] You've been listening to the Solspace Podcast.