Chris Gilbert: Creative Developer

All right, here we are, another episode of Escape Velocity. Today, I'm talking with Chris Gilbert. He is an incredibly sharp software engineer with over a decade of experience building performant, interactive and accessible websites and applications. Beyond just development, Chris is a deep thinker about the evolution of technology, the impact of AI ai on creative and technical workflows, and how companies, especially those in traditional product or service-oriented industries, should be rethinking their approach to internal tool development and process. He spent a lot of time thinking about the shifting role of agencies, developers, and how AI is fundamentally reshaping our ability to create. So we're going to dig in on all of that today. Chris, thanks so much for joining me. um Yeah, i'm like I'm excited to dive in.

Thanks for having me. It sounds good. Let's dive in.

And ah you know we've been kind of having our own convos about a lot of this stuff. um But you know what started happening when when you as a developer were it's just sort of like, hold hold on a second. This is a game changer.

Oh man. Probably the beginning of, well, beginning of Q2 last year, stuff started really taking off 2024. Yeah, just seeing some of the tools that were starting to pop up and what was possible versus traditional ways of thinking was just mind-blowing. And and you know earlier on, there was a lot of doubt that was cast on AI, you know especially from the developer crowd. you know Knowing the technology that is working behind the scenes to make all of this possible,

Right.

it does kind of become hard to believe that it's going to be able to do all these things that we've had these all moments from, right?

and Yeah.

What LLMs are doing behind the scenes is is figuring out what the next word should be, basically, and tokenizing that and tokenizing phrases and figuring out which tokens are more likely to come next. So it's not really understanding these these deep concepts, but the language model and pre-training, it's coming up with um Answers that sound great and oftentimes in coding context are great and allow you to skip forward. Yeah.

Yeah, so I think it's, um you know, with AI, I think we we reached this point where, and i'm I think this was the same with code as it was with copy or content. It was like 80% really good, like surprisingly good. And then there'd be 20% that was like either wrong or crazy or, you know, like just so off

Yeah.

but it was always sort of, it was like, okay, well, if I pull that out, I still have 80% of pretty workable, good stuff in here um that I can use to speed up my process, right? yeah Whether I'm using it to, so from a developer's perspective,

Absolutely.

Did you find it first more interesting to have it like review code that you had written, or were you were you going in and like asking it to write out code and comparing it to what you might have done? Kind of wondering what you're what your initial like experimentation process was.

I'm trying to think back because the space has moved so fast, so my process is iterated so much that thinking back that far is kind of hard now.

Yeah.

um Yeah, I think diving in, my first experience diving into it was... really you know It it had made some progress, some solid progress before that point. so I didn't really get into a lot of the GitHub co-pilot stuff early on, but dove straight into Cursor, an a AI-powered editor. um and was They didn't have a composer at that time. It was just the chat. right so it was You you would go into a file and you can start chatting with it or query the agent to build directly within the file. And it was making these updates or like adding functionality and then having tab autocompletions where I remember writing a a script or a script tag onto the page and then having some functionality pop into autocomplete probably would have taken about 15 minutes to complete it and knew it instantly what I was about to do and just press tab and boom there it's done.

Mm-hmm.

um That was kind of like a ah mind blowing moment for me.

And that's interesting, right? Because you're able to tab along and sort of see it in real time instead of hit submit, and it just spits out a whole bunch of stuff. So that seems like that must have been a very, like, building on, like, wow, as you're watching it complete things and tabbing through them.

Yeah, and at that point, still even you know there's so much that has happened since then where now, but going back into that moment in time, you know it was really crucial to still know what you were talking about because

yeah Yeah.

I break down, you said 80-20. I break it down into 60-25-15. 60% of the time, what I'm getting is is great. Move on to the next thing, good enough. 25% of the time, it needs a little massaging, which typically looks like interacting with the the chat and cursor and saying like, I see what you did there, but let's do this a little bit differently. Use this approach instead of that one. massaging it into something that's closer to what I need.

Okay.

And 15% of the time, it's like completely out of left field. And no just scrap that, start from scratch. My prompt was not good enough. Let's let's start over and see how we can ah tackle this early.

Yeah. So you mentioned how quickly everything has changed. So today, if you were a developer, and you've kind of, ah for whatever reasons, you know not really dabbled in it or gotten involved, like where does a developer get started to to see if if you know to to see the power, to see the opportunity, um and to make good use of these tools?

Yeah, it's a good question. I would say the tools that I would recommend jumping into right away, if you haven't already, I mean, everyone's on chat GBT, you know, like everyone's there at this point in time. If you haven't, like, just start asking it questions and playing around with it, you've got to get in and kick the tires. Specifically with with a developer focus, like, ask it some coding questions, have it spin up an application for something, see how you can interact with it and say like, well, change this to use Tailwind CSS instead of what you're what what it gives you. um Change it to utilize a different library. See what that's giving you. That's very interesting and and you should have that experience and know what's possible in that context. But then also start working with AI editors. It's a big one. um you know There's only so much that you can do with an extension on VS Code and a lot of these applications have realized that so they forked VS Code because it's under MIT license and they have these these editors that are ah have AI baked in essentially into the editing editing experience. So they can do a lot more. um Cursor is a really big one. Windsurf is another.

Yeah.

um Start working with those and start playing with them. I'd recommend probably just go to YouTube and you'll find some videos on on what's what and how to work with it and all that kind of stuff.

Yeah, I was just going to say, I know there's been a lot of sharing on X, on YouTube, but um even in articles on Medium, where other developers are are sharing with you know how they're getting started or what they're experimenting around with. And I i do think that's the best way to learn. And what's great is that it's also exciting and new and um that there isn't any protocol just yet. um there's There's no you know dummy's guide to writing code with AI, because it's just changing too quickly. So there are a lot of people sharing stuff, which is great.

yeah It's definitely in a very fast space right now. you know it's it's It's moving along very quickly. um I do expect that we're going to see a lot of back and forth between the teams. I think right now, the big ones is is GitHub, Copilot, Cursor, and Windsurf, and they're all competing with each other, and they're all just like back and forth with features. um So and until we see someone come up with some IP that's unique that they can defend, you know I think that it's just going to be a lot of back and forth, and it probably makes sense to pick one and stick with it.

Yeah. i mean It sort of reminds me of ah you know the the browser wars, or um yeah you've got chat GPT, you've got Claude, you've got ah whatever the Bing one is. ah yeah there's just There's an assortment of a very similar um chat GPT-like uh, applications to engage with and each one's a little bit different and nuanced. So I do think it's just about trying them, finding one that just feels good to you. You know, it's got the right vibe. It's, it's kind of angled in the direction you want to go in and then, yeah, just stick in because they're all upgrading so quickly. They're, they're going to be keeping up with each other.

Yeah. Yeah. And you touched on this. It does feel like those days and it's very like the the excitement you can feel is like those days, like endless possibilities.

Yeah.

And like, you know, it felt like we were kind of like in this period where we were kind of stuck and just kind of accepting that like it is what it is now.

yeah

So it's really exciting to see all of this innovation happening and feeling like, oh, it's like to the possibilities are endless again, you know?

Yes, I was just telling someone that this is the first, this reminds me of when I just got started in the web world of web and, you know, interactive design, digital design, whatever you want to call it, um building websites, you know, putting stuff out on the web. And ah that was like, 27 years ago.

yeah

And, you know, the iPhone like that flash made things exciting for a while. ah The iPhone definitely made things exciting for a while and and then um there was some standardization and then accessibility became really important and that kind of brought some focus. But it was all just kind of locked. It started to feel a little bit stale and a little bit like ah rinse and repeat, rinse and repeat.

Yeah.

um And now I think, yeah, the floodgates have opened.

Yeah.

So I mean, how do you see this changing the world of websites as we' you know we've spent a lot of time in that world?

in ways that I'm sure we can't even imagine right now. you know um I think it's fit's it's grown more important recently to, when you're talking about AI, discern the difference between the current state of affairs and AGI. um you know And AGI, I'm less interested in talking about, just because when it truly gets here by the definition that I think most people should hold it to, it's just game over.

Right, right.

you know like it's

Now, when you say that, when you say game over, like explain to me what is in Chris's head when he's thinking game over, what's happening with these AGIs.

ah Um, AI powered agents are performing basically every, every knowledge work job. Um, they are incorporated into robotics that are handling physical labor. There is very little leftover in terms of like typical, like standard jobs, you know, even white collar stuff is like just taken over.

Yeah.

And what does that look like? Where do humans fit in? What is valued? Um, It's big question mark right now, you know, and it's, it's, it's scary.

yeah

And I don't think we're prepared for it. Um, Down the same rabbit hole. Um, there's a great episode of the Tim Ferriss podcast where they touch up with the Chris Zaka where he touches it on this a little bit. And he's just like, yeah, I don't know where we go from here.

yeah

Um, do, do, do, do.

Yeah, i' but it would upend everything, right? If you've got robots doing everything, um is there even the need?

Yeah. Yeah.

I mean, do we just sit around our houses and get a ah handout income from the government so we can pay for our food and basic necessities?

Right. Basically. Yeah. Yeah. Are we headed towards the world where it's Wally and we're just all on floating chairs and robots are giving us everything that we need or is is there some other future?

Right. Well, here's the thing.

that

A lot of things that we do today are because we're trying to address problems. you know we're try and And a lot of that's in support of capitalism, right? So we're doing for doing marketing, um brand, a lot of web work. it's It's to get your message out because you're trying to expand your awareness to get more people to support you or um buy your brand or support your cause or whatever it is. those those needs aren't going to just go away. um You know, there's still going to be people out there that are trying to, you know, cure cancer or, um you know, come up with cleaner energy um who are trying, you know, I mean, is AI going to solve all those problems?

Well, those are, it's interesting. I think it might solve a lot of them. You know, it's interesting that you touch on those two points. Something that Chris mentions in that podcast is he thinks we are single digit years away from unlimited clean energy and that there are AIs that are now working to solve um basically a biomedical, like

Mm hmm.

medicine and and solving those problems and thinking about them from angles that humans haven't been able to think yet, like approaching it from a brute force standpoint, trying combinations at speeds that would never be realistic for us to try, where then they can say like, found this, let's go try it.

Right. Yeah.

You know, so it is kind of crazy. Like, I think we can solve a lot.

Yeah, so with the scale of AGI, just just to give it some perspective, right? um If you've got an AI right now that can run through a ton of data and just, like you said, brute force, look at a a a problem like a disease and it can run a lot of different sequences and approaches to see how it might impact that um that medical you know situation. with AGIs, you're essentially going to have unlimited ah simulated test subjects as well as testers, right? So I mean, doesn't it just kind of?

Yeah, I would say like right now, any any any AI that's um in that space is likely targeted and running in a very specific area and task.

Right. Right.

With AGI, it's going to have all of the ancillary and periphery knowledge to approach from all different kinds of angles.

Right. And each AGI that hits a roadblock can spin up additional AGIs to go tackle that particular problem, right? So they can kind of self-populate to a degree I can't even fathom.

yeah Yeah, I it's i would i would kind of like think of AGI as the central entity that it could then theoretically spawn out instances of other AIs, but it would be the the central ah general intelligence that would have all of the context awareness of basically everything, which is insane to think about.

Right. it is um It is. It absolutely will make your blun your brain explode. And I do think it's it's too much for a lot of people to even think about. um

Yeah, I mean, the way that Chris talks about this you know the the this, this computer can now do something better, faster, and cheaper than most humans.

i Right.

If we talk about like, so it's important to also look at progress, right? So one thing I want to talk about is the progress of AI and specifically the Arc AGI benchmark test that we're running all these models up against. There is a model that our benchmark that we're starting to use that is more advanced and harder to score on. But in terms of what's been used to date, it's mostly the Arc AGI test. So GPT-2 in 2019, 0%. It's not until 2023 that we get GBT4 that scores 2%. GBT40,

Yeah.

2024, 5%. The 01 Preview, 2024, 21%. 01 Tuned High, 32%. 01 Pro, 50%. Now they just came out with 03 and it's not out publicly yet, but Tuned Low, 76%. Tuned High, 87%. We're at 87% where I think the average PhD human will score 85. And all of that progress came over 12 months from 5% up to 87%, 12 months.

It's a lot.

so Yeah, this this exponential curve, this idea that like we're looking over the horizon of all of past progress, and what we don't realize is that we turn around, our nose is pressed against the exponential curve that's about to take off. you know it's

Yeah, it's like Moore's law times um with the major you know x to the 100 degree in terms of the amount of of speed here, so we can't we we aren't going to be able to keep up. so at what i mean We can only change as fast as we can change, so what happens when the AI is just more capable than then we are of adopting it?

Right. Yeah, it's i mean it's a great question. I don't think it's a chicken and egg situation, right?

Yeah.

like I was having a good conversation with some people about this recently where, like what do we do in terms of school? you know um The comment was made, why are we even sending kids to school anymore? you know Where are they learning? and I kind of take the stance that what you're learning in elementary and middle school is not going to change. Those foundational skills are so important, learning to learn basically, that's not going to change.

Yeah.

But for the rest of it, looking beyond that, we're kind of in this chicken and egg situation where we need to know how this is going to shake out to know what the jobs are going to look like so we can prepare the next generation for what those jobs are going to look like.

Yeah.

It's going to be a little messy um in the in the middle of time.

Well, i you know, like you said, the knowledge workers, the, um, and And a lot of the things that can be done through robotics are gonna be ah you know factory work, assembly work, um production line work, that's all gonna change. But I mean, I know there're they're working on robot you know robots that can be like in your house doing you know doing your laundry, um preparing food for you, like basically being sort of your household assistant. I don't know how, how near that is to being reality, but you know, that's all going to be at this expense. And so how much is this gonna, um, really supercharge this division between the haves and the have nots in terms of our our class structure and our are economic breakdown in our in our country and in the world. Because the people who are gonna be losing their jobs and not able to continue to have sort of forward momentum and growth in their professional lives and income levels are not gonna be able to afford a lot of the luxuries that the AI is gonna be able to provide

Yeah.

um But yeah, then what are people going to even be needing to do with their time if they're not sort of overseeing the AIs and designing more things for the AIs to do?

Yeah.

um

It's a great question, and I'm not sure that we know yet the answer. you know it's it's It's going to swallow such a larger proportion that it's you know it's not just the blue collar workers. It's the white collar jobs that are at risk now, too.

Yeah.

It's going to it's going to require a lot of, I don't know, rethinking, for lack of a better word, of how to approach this from a not just a national standpoint, but a global human standpoint.

Yeah.

How do we how do we handle this? um with If and when AGI does become a thing that exists without that without a doubt without question, I think it's more going to be a question of, at least earlier on, does economically make economically makes sense? um you know if you

Right. I mean, I know people for a while were saying it's going to take so much energy to run this stuff, but now you've got DeepSeek coming out and sort of being like, maybe it's not going to take so much ah power.

Yeah, I don't know. I'm torn. I want to want to see where where that settles. you know um We have open AI coming out and saying, well, they they used our training data.

Yeah. Yeah.

um Yeah, so we'll see where that comes out. I'm not sure. We'll see. We'll see. But you know the O3 model that is scoring at 87%, they're separating that into tuned low and tuned high from what I've been able to find online. Tuned low is basically $20 per prompt on that model.

Wow.

Tuned high is $2,000 plus per prompt. So that cost, assuming that AGI is going to be more resource intensive than that, right are we able to get the efficiency up to where it's economical for everyday people or even companies, large companies that are used to spending millions if not billions of dollars

yeah

per year and in revenue.

Yeah, you've got to really know it's going to be worth it or and it's not going to just spin out of control and and go cost you millions of dollars and come back with a bunch of hallucinations, right?

Right, exactly, exactly. oh So, yeah, I mean, there's definitely an economic like question there in terms of is it, when it becomes technically viable, is it economically economically viable?

yeah Yeah, so I mean, you know you're you're talking about, and and I love that, like thinking about where we're at in maybe single digit years from now.

Oh yeah.

and But I mean, there are still people that are like blown away by mid-journey and seeing a a generative AI you know produce an image. um Do you feel like this is all just sort of like shiny objects on the road that will soon be forgotten? um Or do you think that there is a there is a role where we continue to use these tools for their efficiency, their collaborative nature, their ability to iterate um that doesn't just get you know swallowed by the black hole of AGIs?

You know, I said at the beginning that AGI isn't really something that I'm... don't know. I think that we're headed towards AGI. But I don't think that... I've yet to see a model, let me put it this way, I've yet to see a model that works so well that I'm not... I'm stumbling on myself here. I need to see more.

Yeah.

I need to see more from on.

I mean, they haven't proven it yet. They haven't proven it yet. I mean, I understand that they've been talking about the hypothesis of it and promising it's coming, you know, and it's always with bated breath, like, Oh, sooner than we said last time, you know, here it comes.

Yeah, yeah. Yeah. And you know, open AI is saying that they know how to, they think they know how to develop AGI.

um

And they've said with Microsoft that they're setting, you know, I disagree with this. they've They've said that they will know that they've reached AGI by the point when they reach a hundred billion dollars in revenue.

Yeah.

I think that's completely ridiculous. Like that's not merit based at all. That's just, we've made money.

yeah

So I do think that we have the tools, the skills and the talent to go build it. I have yet to see something that I mean, obviously not working as well as AGI. It's not here yet.

Yeah.

um But <unk> I've yet to see a model that kind of is like, oh, it's around the corner or it's like eminently coming because this is so close.

yeah Yeah, I mean, there's a lot of there's a lot of um pearl clutching happening about AGI right now, and and rightfully so, right?

you know

I mean, I think it opens up sort of a Pandora's box of questions. But if we put that aside and we talk about, OK, well, what's working well today then that's changed pretty quickly um and improved pretty quickly?

Yeah, for sure.

Yeah.

I think that the tools will continue to get better. They will continue to get better. And you know from from the same standpoint of is is AGI economical when it does come around, I'm not sure that it will automatically swallow all of those tools, you know even though it might be able to do a better job even using them itself and not being human directed.

Right.

I think there's a ah ah large possibility that we end up using them ourselves for a while, you know, at least until the the efficiency can make it economical.

Yeah. Yeah.

So, you know, what does this mean for Dev? Well, first and foremost, I think it means that you're just inherently less productive if you're using it, not using it, and your peers are.

Yeah. Yeah.

um From an agency perspective, not using AI over the short to midterm means that you're going to start getting under bid.

Yeah.

you know You're going to get undercut by teams that are able to use these efficiencies and then get work done quicker.

yeah

um And then in terms of like, what do you do with that time? Where do you place that time? How do you think about billing? How do you think about all these various dynamics? I think that you know it's up to each agency to decide how they want to use that time.

Yeah, I mean, I think it's it's a perfect time too, because it just, it needed to be disrupted.

but

It was too it was too locked into these processes that were like, you know, in these black boxes. And, you know, who are who are you to doubt that, ah you know, that a dev is going to take this long and and you need another dev to let you know if it was good, well done dev work, you know? And same thing with creative. It's just all kind of, um it can it can get very subjective. And I do think there's going to be some playing fields leveled by all of this, but the efficiency and the collaborative aspect of these tools, and I would say both on the design and development side of it, um should benefit everyone. Both the client in terms of the the speed and cost of working with other teams um should be dropping, but it's still going to take ah the right team with the right approach to get quality. Right. I mean, you're still going to have people that can very quickly turn out shit and you're going to have people that can very quickly turn out really good stuff.

Absolutely. Yeah. I mean, you you're only, you're only producing the bar that you're holding it to, right?

Right.

So if if you have super high standards, then you're going to hold it to the super high standard and you're going to accomplish that level of work. If you have low standards, you can, you can produce crap. Sure. Like probably really fast.

Yeah.

um

Well, it also, I think it's opening the door in the same kind of way that um that the digital camera did, or the the early like website builders did, um or the way that desktop computers did to the publishing, you know the print publishing world. It's like the tools are going to let more people dabble.

Uh-huh.

and and be able to spin up something of their own to be like, well, how about this?

For sure.

Because now I don't need to know Photoshop to go in and create a visual, um some different layouts. If I'm just trying to like capture a mood, I can take language and let that language either generate code that could impact a browser or a piece of software. um And I can do the same with visuals and with video ah increasingly so. So I think it just lets a lot more people into the conversation and it forces everything to compress because you've you've got to care more about the communication and the teamwork and the collaboration because you can't hold firm to that waterfall black box handed to black box kind of process. Cause there's no more, that doesn't make any sense. Um, and people are going to break that down, whether you want them to or not, which I, I love, you know, I think that's, that's a great creative change to have happening.

Agreed. Yeah.

But I think that a lot of people are, Um, we'll point towards the lack of tested, you know, tried and tested validity of some of this stuff and the hallucinations and the, the mistakes and sort of say, Oh, it's just amateur hour. And like anyone who's using this stuff, it's just, there's, you know, they're fakers, they're spammers, they're whatever.

I mean, and to those people, good luck. Like, you you're going to fall behind so fast. So, like, what's what's new today, right?

yeah.

Recently, Cursor, and I believe Windsurf has something similar, released their Composer, which is essentially a self-driven agent. um I've been working on a project just to kick the tires with these things. And I said, but let me go create a project scoping application.

yeah.

You know, something that I'm familiar with intimately is a problem and let me solve some problems. I spent probably four hours yesterday just getting the idea out of my head. So looking at where are the data types? What are the actions that users can take? Where are the user roles? That kind of stuff. And then what does this look like from a template or a view standpoint? You know, what what different views does this application have? And I just got it all out there. And I optimized it, of course, through chat GBT for clarity and all that kind of stuff.

Yeah.

I was able to then today feed these directions into cursor composer and watch it not just spin up one file, but create a whole directory structure, spin up every single file in the application, write all of the code and solve its own bugs coming back to me with, okay, I'm done. Here's what I did.

Wow.

A to Z, just start to finish. Now, were there still bugs in the files? Absolutely. But I think for what I did in five hours, that probably was a good month and a half, maybe two months of work if you did it by hand, a hundred percent.

Yeah.

That's just crazy.

Yeah.

You know, um does it need further optimization? Absolutely.

Yeah.

You know, I'll need to go in and make sure that the right user test and all that kind of stuff. That's just insane. Like we're at this point now where the the functionality, the the layout even is just there for prompting. You know, it's it's not this manual.

Yeah.

It used to be like bricklaying almost, you know, where you like sit down and like write the markup and, you know, like, like craftsman level, like just chiseling away on, on some marble, you know, and.

I mean, it was a lot of bricklaying. The whole process to to build software or websites or anything digital was a very ah kind of, now that now that we think about what's possible, it does feel like archaic to some extent.

Yeah, but it's, it's, it's, it was a crazy experience just watching that happen.

I'm sure.

I was just like blown away. You know, you're sitting there watching it and it's just like, it tells you what it's doing and what it's thinking about currently what it's doing.

And that's in cursor.

And as it. Yeah.

Yeah.

Yep. And as it is spinning up files, it's hitting errors. It's aware of those errors. It's going to fixing those errors before reading, even told, you know, and then

Yeah.

you end up with a thing that's like ready to go. And if you hit a problem, it just fixes it. um

Yeah, I think that's really interesting that cursor will show its thinking. um I know it's been really interesting to to tell chat GPT, show me your thinking, walk me, don't just make changes.

Yeah.

Um, but along the same lines with, with custom GPTs, I'm like, why would you ever be working with a client and not be creating a custom, a variety of custom GPTs just for that client work? Um, because you're gathering all this information, you're having conversations, you know, if you're recording them, You have transcripts, you have notes, you have documents and the information that the client's giving you, you've got stuff that you're going to go out and find all your own research.

yeah Yeah. And you can give that custom GPT web access, and it can go out and search the web and also find stuff, right?

Yeah. Yeah.

So if there's like a knowledge gap, it can go find that. It can search the web even. In terms of of using code this way, you know there's pros and cons, right? And I do think that it is still Like you mentioned, the senior developers are more suitable for for this role.

Right.

um You still need to know what you're doing. You need to know what's going on. So I would say a couple of things. If you don't understand the code that's being generated, that's a problem. so but But it's not at that end.

Yeah.

Hope is not lost.

Yes.

You can you can ask the the AI in cursor, the chat, like explain this file, right?

So it can teach you while it's working with you and for you.

If you don't know what's going on, Yeah, absolutely. Not at the same time that it's generated code. But if you go into files after the fact, you can say, explain line by line what's going on.

Mm-hmm.

And it's going to tell you.

Right.

So it's also a great way to learn. I think that's going to separate a lot of people using this.

Yeah, I mean.

There's the people that are just like, want this quick thing. And it's like, do this thing for me. I don't want to be bothered. And they're going to take it verbatim.

Well, look. i Right, how long has ah you know how long have Harvard and Yale been offering free courses? You know um you can get a free education in pretty much whatever you want if you just go find the resources. um But people do, you know they're lazy, they're intimidated, um they don't necessarily want to have to learn it, they just want the quick results.

Yeah.

How do you think that's going to play out though with people who who need those entry-level jobs to have that experience. Maybe they went to school, got a computer science degree, are fairly familiar, but they're not going to get hired into senior-level positions. like What is this going to do to that entry-level, junior-level, dev position? And I would say, you know across the board on on anything that AI is touching.

it's It's interesting. It's one of those areas where I'd say we don't have the answers quite yet.

Yeah.

One area that does occur to me is you know as as these systems get more advanced, I could see there being a lot of value for someone that has a expertise in an area outside of Dev where there is low hanging fruit for creating applications inside of that niche.

Yeah.

and and breaking in that way. um But from a pure programmer standpoint, I think it's it's going to get difficult, at least in the mere turn.

Yeah, i I mean, this might be wishful thinking on my part um and and just being optimistic. But you know the one thing that the AI is not ever going to have is its own lived experiences. Right. And so if you went to school to get a computer science degree, but you also spent some time over in another country and you, um, you know, your summer jobs for a while were as a chef and, or, you know, you worked in a kitchen, um, you're going to bring your own unique And that's like that's the creativity, right? But it's you're gonna bring your own unique combination of lived experiences to the table as you collaborate with these powerful tools. And it's gonna change the way that you prompt them, the way that you communicate with them, the way that you interpret what other people are saying to you, and then how you're using the tools. And so I i think it leaves a lot of room for those things to shine in ways that they should anyway.

Yeah.

um And that might be the magic combo of, like, if you're just getting started, like, yes, learn learn the foundational stuff, understand the code, but also bring your own unique perspective to it.

the Yeah. bring Bring creativity, communication and empathy. You know, those are going to be highly sought after valuable things, I think in this world where the the technical skills.

Yeah. Yeah. we don't we don't We're not going to need more. Yeah, grunt coders that um when the AI can just do it all.

Right. Right. Exactly. I would say understand the tech stack for sure, and have a general idea of best practices when you go into something. That's that's huge when you're using something like cursor composer, you know, like, you know, it really comes down to experience and just getting your hands dirty and jump into it.

where where do Where do people, where can people learn about that? mean you know

And, you know, there's no better place to learn than this right now, you know, like jump into cursor and just have it spin for project, but then have it explain to you everything like until you understand it and make edits and figure out how those edits play out and the impacts they have the downstream impacts and you know, there's

Yeah. Yeah, I think that I've seen a lot of people go into an AI environment and they'll just type one thing, one sentence, and they'll it'll spit something out. And then they'll go, oh, that's cool. But I've seen people who will spend a day communicating back and forth with the AI, um iterating, adjusting, pivoting, adding, you know re restarting at at times. but the you know and youve and you've got to i think really have your You've got to be able to curate what it comes back to you with so that you don't get overwhelmed by the volume of data or visuals or whatever, because it can spit out so much at you so quickly.

Yes.

So you've got to be so discerning about which ways you want to tweak it and have it go. um But it's you got to stick with it. It's not it's not a it's not ah quick fix um to get it to really do its best. you know it It does take some work, but again, spend that day and you you might have something that would have normally taken a month.

Right. And depending on your personal learning method preference, you know, you might get a lot out of that, but yeah, there's, there's no shortcut, right? There's no shortcut for experience. So just getting that experience in is huge. And then, you know. I wish that this existed when I was learning, quite honestly, because I didn't have this this thing that I could just ask endless questions of. you know right I had the questions, but like you Stack Overflow, right?

Yeah.

like like Searching Google um was was the extent of it. so yeah If you have questions, like you have this unlimited resource that's never going to be like, oh, I've got to stop asking me questions. and

here's the Here's a, you bring up an interesting topic, right? Search. Um, I don't find myself going to search nearly as often as I find myself going to chat GPT these days.

Yeah.

Um, yeah.

Yeah. Or even using stuff like perplexity, which is like much better at just general search. I find, especially if you have like a specific thing that you're looking for, if you're looking for a specific product or you want to find the best X or, you know, like stuff that might involve like a low level of research.

Yeah. Yes. Yeah.

it's It's so much better.

perplexity amazing

you know It's not just like, oh, here are five articles that say they've got the best travel spots. you know It's, oh, doing some research from that data and then coming back to you with like, okay, here's what people say is the best.

Right.

um Another one is Reddit just released a AI platform called Answers.

Oh.

So if you've ever like gone down a Reddit rabbit hole trying to find like the perfect, it it has all of their,

Yeah. Right.

data like basically trained on. So yeah, it can just like summarize you a good answer.

So for for people that spent you know a lot of time and money worried about SEL, so they'd show up in search results in Google, um is has the game changed now like to where you really need to think about how the AI is going to summarize your content or weave your stuff into its answers or make sure your content is on Reddit? um

It's a good, yeah.

It just seems like the findability for for things is changing.

Yeah, for sure. So there was, so initially SEO was like, you know, keywords and all of that jazz, right?

Yep.

And then it shifted. And I think this is a shift in the same direction, but even further, right? Like content is kingdom. The more relevant your content, the more likely it is to get pulled to the top. I think that's just underscored even more with this. If you have relevant content, the AI should scrape that and then be aware of it.

Yeah, it's um it's really interesting in Google now to see the the AI summarization answers above the actual search results. um And that's not sponsored.

Right.

Nobody's paying for that yet. um But it just opens up such a can of worms.

right Yeah. And I think they actually need to go a little bit further down the path of what perplexity is doing. If they're going to stay competitive, just offering up these like summaries and not doing any kind of research or digging when it's being prompted with a literal question just seems a little bit like a fail.

Yeah, and well, I'm sure that they are all racing to improve.

Yes.

It's quite the race. um ah what What are your thoughts?

For sure.

I mean, I know you're you're a fan of Apple products. um I've just been so disappointed with the Apple Intelligence and Siri.

so yes Yeah.

I mean, come on, Apple. like

Yeah, same, same, same. And I know they just hired a new, new role to, or put someone new in the role to see overall that.

Yeah. yeah

So maybe we'll see some progress here. I don't know. It seems like they kind of like went AI light and then pulled in open AI as like a fallback, you know?

um Yeah.

Um. I don't know, Apple has traditionally though done impressive things when they've been kind of like put into the underdog position.

Yeah, and they're maybe second or third to the game.

You know, they don't, they don't, they don't respond well to being in that position and they really, really go all out to kind of not be when they've been put it in that position.

Yeah.

So I'm curious to see what they respond with because it's very clear that Apple intelligence is, is not living up to the competition.

is a big nothing burger. Yeah, I mean, it's like, oh, there's a lot of hype for nothing.

Yeah, yeah, yeah. yeah Oh, yeah.

um But you know, back to your, your, you know, thoughts on like, Will we even be using you know monitors and screens and and that sort of thing? and I thought Apple Vision Pro was ah was a really interesting look at moving in that direction with them with VR and AR and all of that. But even just the need for interfaces, the way that we've typically been thinking about them, um that could all really change in the in the near future.

Yeah. Yeah, I think you know it depends on... Can we get a form factor and price?

Yeah.

For hardware that makes sense for mainstream users, you know, I would love Some glasses similar to like the the form you're wearing now where I could get some AR for that, you know I'm not sure if if sitting at my desk with goggles is is exactly where I'm at right now, but I know some people to do it.

Yeah. yeah

So it's it's definitely coming, you know, I think they're only gonna get slimmer and better But yeah, once that's there, like what happens? you know I think computers and monitors disappear from desks first. And then if you're thinking about a VR context, like what do websites start looking like? you know A lot of the flat 2D representations, I wouldn't say they fall flat, but they're not limitations anymore.

yeah I think one of the biggest changes is is how we've historically just structured websites and we've kind of locked into this very formulaic and very, um you know, it's it's like a library system or ah ah an index of of a book. I don't think people are engaging with information the same way that they have been.

Yeah.

And I think that that will continue to change in really interesting ways. So I think that the whole notion of like information architecture and navigation is is is just going to feel like, again, like laying bricks um when that is not how we are consuming information.

Yeah.

Everything is getting a lot more conversational and and run through these these chat GPTs and things like that. So I'm excited to see what changes there.

Yeah, for sure. um You know, I think that a couple of things come to mind. One is that I think the the experience is going to become a lot more cinematic and not just a flat, you know, screen, basically.

Yeah.

I imagine, you know, What comes to mind is like the the NVA website and like being like surrounded and immersed in a hearing stadium you know with that energy and how that hits you and versus how a website can hit you.

Yeah. Yeah.

you know it's It's night and day, right? Suddenly you're on the ground floor you know looking at this this gigantic stadium pool of people. I get chills just thinking about it. you know

I know if they if they can get the form factor right on that and the price point, who's going to want the screen over that?

Right. I think, you know, we might see something in the way of a a custom navigation, but I think voice is going to go a long way there.

Mm hmm.

Not many people, myself included, like to use voice for navigation, but I think that is the medium that it's going to take to get that more mainstream.

Yeah, I know it's been interesting to watch like cars um play around with gesture control and voice control. you know ah Cars have been a great, automobile industry has been a great place to watch that kind of evolve and succeed or not succeed.

Yeah.

And and at this point, you know i I do kind of rely on my my Alexa to to do certain things. um But it it does seem like it's just so disruptive and kind of clunky at this point.

Yeah.

So how do you, you know, how do you separate like normal life conversation with the conversation that's meant to engage with an AI or an interface um without learning a whole new language or resorting to you know gestures?

Right.

um

and And maybe gestures do also play a big part, you know, I could see that as well.

o Yeah. And I, you know, I mean, so many people use the voice to text features on their phones now. Um, I think we're getting a lot more, it's like, I'd rather FaceTime and talk then, then call.

Yeah.

Um, but there's still, you know, there's still some people who are like, I'm going to text that's, you know, and there's still email. So there's still letters in the mail. So we've got it all.

Yeah.

We've all got it all kind of rapidly spreading out, and um some of it's going to condense, and some of it's going to cut off. But it's interesting times.

Yeah, yeah it's it's it's interesting to think where this could go. know What is it going to look like when we don't have any desks and it is all just AR and VR?

Yeah.

and I don't know. i I'm mostly curious to explore and and brainstorm on what we've traditionally thought of as websites look like. And are they even called websites? But like what does that experience look like when it has the level of standardization that we have for websites today?

Yes. Now, you know to be devil's advocate here, people have been predicting this shift for the last 30 years, right?

Yeah, yeah.

And it has not come to fruition. And we're basically now saying, well, because of AI, that shift is finally here and it's going to happen quickly now because of the power of AI. um But it's, you know, I think that's rightfully like the jury is out.

Yeah.

So we will have to wait and see. um It will be it'll be strange if the world of websites goes away, um but I'm excited to see what might what might replace it.

Same.

Yeah.

Same here. Yeah. Back onto AI-powered Dev, I have a couple of other things to to mention.

Yeah.

um One thing is that I feel like there's a huge opportunity and a lot of low-hanging fruit for internal AI-powered applications at companies.

Yeah, go for it. hey

you know and I see two primary use cases for this. One is the product and deliverables focus, and the other is internal processes. and you know there's a lot There are a lot of services right now that you can go out and subscribe to to handle a lot of things. But I think now more than ever, there's the use case. like Your dollar stretches so much further for internal app development that it makes so much sense. I mean, there's there are companies that I'm sure

Are you talking about like proposal five versus building your own proposal writing?

Right. And I bet if you went to a company, they could probably come up with a long list of oblig applications that they would love to have, but just economically cost-wise doesn't make sense to go build that.

Yeah.

It's, you know, however much that is, it's, it's much more, it's much lower hanging fruit now than it was, especially with, with AI when, when you can spend up a full application and five hours after spending a day coming up with a planning, you know,

Yeah, absolutely. Yeah, I mean, there are a lot of people saying, you know, this is the death of SAS right and now. It's going to be AI as so as a service.

Yeah. Right.

um you You won't go to software anymore because you won't need to.

Yeah. Yeah. Like going to like specific to a company, what can you automate? What repetitive action are you taking? What grunt work is required? Where can you insert AI into time consuming tasks?

Yeah.

How would you organize and break down your path to creating products and deliverables? I mean, how can AI not only perform those tasks, but connect them with agents, right? So it's the same as the cursor composer.

Yeah.

You're not like telling it, go to this thing. Okay, now go do this thing. It's here, do this. And it goes out and just does it all, right?

Yeah. But here's the thing, you are thinking about that. That's what I, every person I talk to now, especially if I'm hiring someone, a young person in particular, I'm like, tell me what you're doing with AI. You know, what are your ideas? Like, what are you playing around with? Because I want people that are thinking about creative ways to use it.

Yeah.

in in everything that they bring to the team you know and the and their own jobs.

Yeah, for sure.

So what else are you thinking about?

The big one, the big other one that I'm thinking about is is data management with AI. We talked a little bit about this, but so many times, in so many cases today, data migrations or content migrations come with data loss. And that might be because of old voice messaging and tone. Or it might be because of messy data that just doesn't translate cleanly.

Yeah.

But with AI, I would make the case that you should hold everything, right? It makes sense to have this repository of like all of the content you've ever generated. And maybe you add to that over time. But it probably also makes sense to take these like snapshots right and have this was 2025. And I need to spin up a page about X, but let me like just ask the AI to do this and use all this historical data and do what you can. And then make that sound as as great as you want it to look, right, or sound. And maybe it needs to be added to, maybe it needs to be modified, edited slightly. You have to input some some new fresh content there, but then you've got that. And that gets cycled back into the repository of that topic X. And it could be organized, organized categorized, you know, so that it's clean and you can go in and explore it in different ways, but, you know,

Yeah, it really adds a lot of dimensions to your data, right? A lot of dimensions.

Right. and But this idea of, you know, I can tell you how many times I've built a website and the holdup is content on the climate.

Yeah, always.

And right. And how how much more efficient would it be to have this repository of data where, you know, maybe it doesn't make sense for Batem, but in a world where an AI and LLM can just say, You can say, adjust the voice and tone of this using.

Yeah, here's the new content strategy.

Right.

Take all the old stuff.

Exactly.

Yeah.

Make it new, make it fresh. Right.

Yeah.

So yeah, I just, it's, it's kind of changed things, I think, you know, and I see, I see a huge opportunity for, um, you know, a product there that hooks into, uh, content management systems, you know,

Yeah, writer writers are not going to like hearing that. Yeah, and I mean, for the dry stuff um or for just the informational stuff, it absolutely makes sense. And don't worry, writers, you know we still need we still need the heart and the human at the forefront and not all just the the bullshit and the flowers that the AI can churn out.

yeah

But yes, in terms of the bulk, the heavy lifting, um the fact checking, the the maximizing your data, Um, it does seem like it's really just having enough people around who have the ideas and know a little bit on about how to do it is, is that's the only limitation.

Yeah, for sure. I mean, it's just going to be such a game changer. I could imagine a product that ties in where you have, you know, a template or view, whatever it is. And you've got 15 fields that you need to populate. And there's just to auto populate, but you know, it's like, take a, take a rough shot at just doing a first messy rough draft of this, you know, and instead of starting from scratch, you're starting from something.

yeah

And that's going to be huge for a lot of people where, I mean, I was looking at a website the other day that launched like months ago and still doesn't have pages with content. I just, it blows my mind that they exist, but like, yeah.

Well, that's because people hold things up Right. It's somebody's job to look at that content and approve the new content or whatever it is And then somebody else's job to move it from one place to the new place um So that can really get slow and tedious.

Yep.

um And you know no one wants to give that up to the AI just yet either. So you know ah it's ah going to be a big control battle as well.

True. Yeah. Yeah. Yes.

Yeah.

Yeah. Yeah. Um, yeah. How are you trusting a computer, right? So.

I know, but I think, I mean, we're trusting them quite a bit already if we really think about it.

True. Yeah.

So it now seems like a silly time to stop, but I do think that's going to be a big issue.

Yeah. Yeah. And this is, that's, that's another big topic of, um, we a shout out to read Hoffman's new book, super agency.

Ah, yes.

So it's, it's all about agency and what AI means for human agency and having control and the feelings that we might have about that and how it's going to shift and things. And, um, and it's It's going to be a crazy new world.

um So that's a good that's a good thing to remind me to say that I'm going to try to get all the things we talked about um referenced on the podcast website, which you can find at helloadeo.com, and look for a shout out to Escape Velocity, or we'll try to add it to the notes on the Spotify or the Apple podcast page as well, because we talked about a lot of things. And I'm sure if anybody's listening in their car, they're like, I can't run. I'm going to crash. you know um So we'll try to get all the all those details out. Chris, this is always fascinating talking to you. So thank you again for giving us your time and your thoughts. And I feel like we should put this in a time capsule and and revisit. um We should revisit this conversation in like a year and see what's, yeah, let's give it a year.

Yeah. yeah

All right. Well, thank you so much.

Thank you so much. Yeah, no, thank you.