Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

Scaling DevTools

Scaling DevTools artwork

Kim Maida on the DevRel Flywheel, AI, and Measuring What Matters

Jul 10, 202648 minAudio

In this episode, Kim Maida joins the podcast to talk about DevRel, AI, developer experience, and measuring what actually matters. We also dive into her developer empowerment flywheel and how Keycard is approaching identity and access for AI agents. Links: • Kim Maida • Keycard • Kim Maida on LinkedIn • Kim Maida on X

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

KimThinking about it in such a narrow scope of what is our awareness value, it does a disservice to DevRel because DevRel is so incredibly broad, and it covers so many things.

JackToday, I'm talking with Kim Maida, a legend in the DevRel and identity space, who's currently at Keycard. She breaks down her DevRel flywheel framework, a five stage cycle that moves developers from initial inspiration all the way to becoming product champions. In my opinion, this is one of the most useful frameworks in DevRel. Enjoy. When it comes to, like, if you join in DevRel, like, how are you, like, learning stuff? Because I feel like it's actually, like I found it's quite hard sometimes without, like, basically, like, dog if you kind of just build examples and stuff, it's like kind of like you you meet someone that's implemented it under a deadline in, this complex project And they just, like, they know things that's really hard to just, like, learn if you're like

KimYeah. That you you can't just, like, put together in a in a toy sandbox. Like, this is, like, real in production enterprise. Yeah. You're like, well, I'd I'd haven't had the chance to build an enterprise project with this yet. So Yeah. Yeah. No. It's the it's always a challenge. And I think that DevRel is just one of those areas where you just kind of constantly have impostor syndrome, right, because of things like that. Unless it's something that you've already used in an enterprise circumstance yourself in the past and then you join, then you do know that pain and know, like, kind of the complications of using a product at that depth. But I I've always kind of just been like, I wanna learn from those people. So tell me, you know, what you're struggling with, and I am not going to turn around and say, oh, you should do this. I'm going to talk to the product team or the engineering team and get answers that I can stand behind and that I know are correct versus just kind of saying, oh, yeah. This is this is what you should do, maybe.

JackYeah. Yeah. So not not like being genuine, like not faking it if you don't know it at all costs. Yeah.

KimYeah. And people people appreciate that. Like, would always rather be told, I don't know, but I'll get back to you over something that someone's not really sure about, and they're just telling me to look like they know what they're doing.

JackYeah. That sorry. That was my side question. So I guess, like, the question I followed from, like, do you care about measuring DevRel is at this stage is, like, what what are the kind of things that you do care about as, like like, pre kind of, like, putting like, launch, I guess? What are the things that you're focusing on?

KimHonestly, it's just putting it out there. Like when a product isn't launched but you use it and then you're really excited about it and you can see the value that it has, it's really just you you can't just sit around and, like, wait for all of the go to market motion to be built up and wait for, like, the self serve to hit the market. Like, you have to just go out and do something. And people will get excited about things and they'll already kind of know what you're talking about before they can actually, like, go and sign up for it or download it or what, you know, what have you. And that is it's just a really exciting time to be a part of a company and a product, like, life cycle.

And this is the first time I've been there at that early of a stage, and I'm really, really enjoying it because you basically get to start from zero. Like, there is no community already because there's no developer users out there. The only developers who use it are, like, our customers and ourselves. And so being able to kind of have that opportunity to do developer experience and to do dog fooding and feedback, like, internally as literally, like, customers here. Like, use like, we say, oh, DevRel, you know, is is playing the part of user zero.

And we say that all the time, but in this case, it's like actually true. Like, you genuinely are user zero. Yeah. And so, you know, like, a new feature will come out and I'll put it, you know, in in my dogfooded project. And and then I'll give feedback and and, you know, the engineers are asking me, yeah, know, can you put this in? And then, you know, let me know if there's anything if you run into anything weird and, you know, I feel like I've have a tangible impact on the product. And that's, you know, it's something that like everybody in DevRel always wants to say that, yes, like the their DevTools feedback does have a tangible impact on the product. And in my experience, as much as we try, it's not always true. Like, in some orgs, you can get great feedback or from other people or you can generate great feedback yourself, and then you kind of throw it over the wall. You you put it like product board or something and and then it just sort of disappears.

Right? Like, especially in a big org. So it's hard to follow-up with people for one thing. Like, if somebody's like, yeah, I told you about, you know, this problem I'm having, and you're like, okay, I'll, you know, I'll let somebody know and get back to you. And then if it just vanishes, you know, maybe it is roadmapped, but, like, nobody comes back and tells you or you don't have the means in the org to just find that out really easily. And being able to, like, actually be user zero and actually just see, like, the answer right away and, you know, be a part of the solution is it's a luxury that, like, I've kind of been dreaming of throughout my DevRel career. Because I think developer experience is so incredibly important and DevRel just is in such a critical position to be able to provide so much value there. And most orgs wanna say, oh, yeah. Like, we definitely do that. Like, yeah, DevRel does that for us.

But it's just not, in my fear, it's not always as true as you'd want it to be. And it's actually true here, like, now at this stage. So it's very that's very exciting for me.

JackYeah. Yeah. Yeah. That's really cool. And maybe could you share a little bit about what you guys are working on because it's pretty interesting?

KimYeah. So we are platform for agentic, credential issuing. So we're a security token service is is kind of the boring term for it. But what we're able to do is issue ephemeral credentials that are scoped to a task for AI agents. So when an AI agent, you know, is told, oh, you know, I want this agent to be able to connect all of these services. I want, you know, these scopes for these specific services. Then when you go and have that agent actually execute a task, you get a super short lived credential that basically just lives for the duration of that task. And so it has, you know, the scopes for that task. It has a lifetime that's you know, it's short so that if it, you know, ever was leaked or something, it would already be expired by the time, you know, anybody got their hands on it. So the policy is basically evaluated before so it knows sort of what it's going to do before it does it, which is something that, like, the industries think at large is really struggling with right now with AI.

It's just giving the keys to the castle to AI agents and then just kind of hoping for the best. And we've seen, you know, we've seen the news articles about about bad things happening because of that. And there's a lot of companies out there trying to solve this problem. But I think part of the issue with the way that various different companies are solving it is because human identity is like a really well established thing. Right? Like identity and access management has been around for a while. There's a lot of people who work in it, and there's a lot of companies who are, you know, very big names and have made their names being experts in human identity. So you have, you know, like, Osteros and your Aktas and and, like, I've actually worked at both of those companies. But this is it's like a new frontier when you have autonomously acting agents or things that are acting on your behalf and you've you want them to be able to access a million different things. But you don't want, like, standing access when you're not doing something.

You don't want that autonomous thing to just be holding on to, like, an overprivileged token. And we our product basically allows you to not have any standing access. And from kind of just my own experience in the I'm industry and working in human identity and things like that, and and then looking at how people are trying to solve this problem right now, it's really become more and more clear to me that we are solving it differently than everybody else's. And that, like, you know, everybody wants to say, oh, yeah. We're differentiated, you know, because of this and that.

But the more I see other companies trying to solve this problem, the more I really am like, oh, we are solving it a different way. And it is better. Like, it is genuinely a better way. So that's it's really exciting to be a part of that. And, you know, to be completely honest, I never really when I went into I'm, like, as a career, I don't think I ever foresaw being on, like, the cutting edge of emerging technologies because I'm moved slowly.

Like, it's just the fact, like, OAuth two has been out for like more than a decade. And like, OAuth 2.1 is still not out yet. So it's like, you know, things kind of crawl along and they get RFCs and they you know, we adopt new standards and then we talk about things for a long time and and slowly gets rolled out. And for as long as OAuth two has been out, there are tons of companies that are doing auth, and they're not standards based yet. Like, they're not even using that yet. So it's it's really not an industry that moves super fast at at a normal you know, under normal circumstances or kind of before this era that we're entering right now. So it was unexpected and exciting for me to have an opportunity to be able to continue to work in identity and access management and security and authorization authentication and also be working with stuff that's, like, coming out right now. And there's, like, immediate problems that need to be solved, and nobody has solved them. And that's just not something you come across in in traditional I'm very often. And I had I had kind of assumed that my career would just kind of be rooted in these sort of slow rolling standards.

And the fact that it's not anymore and that we do get to experiment with brand new things and and, like, actually make, a tangible difference at the cutting edge, that's, you know, that's really, really cool to me because it's not something I expected to be able to do, and I am not gonna lie.

JackYeah. Yeah. Yeah. Yeah. It's yeah. It's funny that that's like the the most sci fi type stuff of like giving agents. How can agents authenticate themselves and stuff? It's like very interesting. I have a question which I don't know if I'm gonna phrase this well, but like when it comes to kind of DevRel dev experience, like now that, you know, there's tools that are designed for agents to be able to work well with agents, like say, for an agent to be able to authenticate yourself in, you know, a system, make payments without getting approval from you for every payment for instance. But then also like there's agents now implementing a lot of the stuff as well as where, you know, the I guess docs have to be like as an example, you you should be thinking about making sure docs are easy to access for agents and stuff like that.

Have you found like it's any of those things are like really different now in like developer experience, developer relations type stuff?

KimIn like DevRel, I don't know specifically that I have yet in one regard because the product that I work on now when I entered the era of AI is not on the market yet, like publicly. So we don't, you know, have a ton of users out there who are trying to use it and struggling because, you know, the because the our LLMs. Text sucks or something. So it it it's I started to feel it though at the last company that I was at, which was Fusionoff. And so that's where I was when this whole kind of revolution started with agent access and agent identity.

And so we we started to think about, you know, like, how do we make our docs basically more accessible for agents, like, as a first party, which which is an interesting shift. But if you think about it, anytime you, like, structure information to make it easier for, like, machines to read, you're also usually helping people because you're giving it more structure. And it it really is like, I'd I've been kind of always attached to people doing things. Right? Like, developer relations isn't about, like, machine relations.

Like, I'm, you know, trying to build a relationship with the humans who are doing things. So I'd always kind of seen it from the human perspective. And it was really, at first, very antithetical to me to think about it from the perspective of, like, humans using these, like, automated tools and, like, setting tools on tasks and and, you know, not necessarily being the ones who are looking at the docs and stuff like that. And I to be truthful, I was a huge skeptic. I was incredibly skeptical of this whole thing.

And, like, when they started talking about vibe coding and stuff and I tried it and I was like, this is garbage. You know, everything it's doing is garbage. You know, I was I was very skeptical. And now being, like, completely immersed in it, I can see the value in it, and I, like, am actually working harder and more than I was before just because I can get so much more done. So I think I kind of maybe got off track from the question that was actually asked, which was around the dev rel for machines almost like in this era.

Right? And I think like it it comes down to a user experience is important no matter who the user is. And, like, yes, machines and people don't necessarily behave exactly the same way. But somewhere at the root of, like, the machine behavior, there is usually a person. And we're not quite to the point where, like we definitely in in getting into the the time where products are being built mostly by AI.

And, like, people are sort of directing the AI, but they're not actually building the product anymore. There's a few companies who have, you know, obviously, like, publicly come out and said that that's what they do now. But I almost feel like in this era, like, human expertise is even more important because you're sort of the last bastion between the AI doing something bad and just like disaster. So if there's not proper guardrails and if you're not asking the agents the right questions and if you don't have the knowledge to identify, like, when it is doing something wrong, when it is going off the rails, when it's hallucinating, or when it's just overcomplication is probably the biggest thing. Like, they really like to make things way more complicated than they should be.

And human knowledge is just like a piece that is so crucial to get anything of quality out of these things right now. And, like, that will probably change as time goes on. Like, you know, when people say, oh, you know, I I had this agent build this fully functional, you know, product, public ready application, and I went to sleep and it had built something and it was perfect in the morning. Like, I don't know what models they're using, but it's not the models I'm using. And so humans are still incredibly important and it it kind of ties back into the developer relations that we're doing now maybe is for human to computer like interaction and like understanding of how your knowledge applies to the tools that you're using.

And yeah. So it's sorry. That was like a super roundabout answer, but it's kind of the the philosophy that I have been operating off of.

JackThis seems like mean, like, Clobbart, like, was built lots of well, OpenClaw, built very with a lot of AI, but by someone who's like incredibly talented and, you know, I felt like all of the big projects are like doing really well. It's not suddenly that like people that weren't also talented before like suddenly it seems like most of it's like the same sort of people are getting like multiplied, like more productive, like

KimAnd I think like software design and architecture is still something that like a machine can do that, but there's a human perspective that machines just don't have. And that is a piece that, like, people you will get a much better product out of it if you are involved in that part of it. Like, even if the machine could write flawless code, you're still talking very tactically. Right? Like, so the strategic part of, like, designing an experience is still something that, you know, from from all of the limited time that I've had trying to build apps, you know, mostly with AI, that's one of the things that, like, they suck at really badly.

So there's there's just a layer that at this time, like, machines just can't do that. And, you know, when the time comes that they can, then we're probably, you know, looking at like Skynet and like The Matrix. Yeah. So

Jackyou know? Okay. I wanna ask you about measuring DevRel. And I know that right now it's like you don't you guys are not necessarily doing that. But you wrote this article on it and you also did you also made an update and were like, I've changed my perspective a little bit on on measuring And so I just wondered how you think about, like, measuring DevRel. Maybe like how you would think about it with Keycard in the future, how if you might be thinking about how you would think about it or how you would have full value in the past.

KimSo, yeah, it was basically, like, based on okay. Theoretically, what how much money is each, like, engagement sort of worth. And it's very hypothetical. Right? Like, we can look at historical data to kind of get an estimate of, you know, how many signs we got if these these many people did this thing, you know. But it's it's all it's all a guess. Right? And I think at the time that I wrote it, like, it was a good framework. Like, it it worked for communicating kind of a a dollar value to leadership about what is DevRel contributing to pipeline. And it really is one of those things though is, like, there's so much more to it than that.

Distilling the value of DevRel into, like, dollars associated with, like, engagements or interactions felt very narrow to me, like, as time kind of went on, especially if you kind of broaden out from just awareness and things like that. Then once you get into, you know, developer experience really deeply or, you know, quality of documentation, helping existing customers, then that model just it doesn't have equivalencies. Right? So I think what I kind of have been leaning on since then is more of just like a flywheel, like, people making their way around a whole cycle from not knowing who you are to becoming like a champion and telling other people to use your product and things like that. So I think that that's really where I had kind of landed on is thinking about it in such a narrow scope of just like, what is our awareness value?

It does a disservice to DevRel as like the mission of DevRel because DevRel is so incredibly broad. And it covers so many things that you're just capturing like a tiny slice of it if you're doing that. But, you know, at at the time and also I think, like, places I worked immediately after where I worked up when I created that framework, which was at Codinary, it still worked, you know, for for years. But I started feeling like I was getting kind of just losing too much because there's there's a lot of value in having, like, a deeper engagement with people. And it was much harder to just be like, oh, that's worth this many dollars.

And also, like, DevRel changed too as far as from a leadership perspective. There's a lot more value recognition around DevRel now than there used to be. Like, when I started doing DevRel, it was like, well, how much money are you making us? How much money are you making us? What's, you know, DevRel's impact on revenue, on ARR, on the bottom line? And, you know, we always struggle to measure it, but maybe we it was because we were just everybody was kind of looking at it through this super narrow lens that doesn't capture what DevRel actually is. And now I think people have a much better understanding of kind of the holistic picture of DevRel. And, you know, like, where I the last couple of companies that I've been at actually cared a lot less about, like, metrics value and that type of thing. They cared more about, like, relationship value and partnerships and, like, early access participants and and that type of thing, which I think recognition of the value of those things apart from just like monetary numbers is something that we just didn't have at the time that I wrote that. And now and now we do in in like a much bigger way.

JackDid they measure when you talked about like relationship value and stuff, were they measuring that as like a thing where it'd be like, you still have like kind of targets whether it's like very specific things or like

KimYeah. It would it was kind of like if a sign up is worth $80, then, you know, a champion who makes a video and then x people watch the video might be worth, you know, $200 or $500 or something because on average, we would get this much and this many sign ups out of it or something like that. So we were still, like, trying to track that stuff, but it was it was very sort of impersonal, I guess, is this maybe the word. And when you have something called developer relations where literally, like, the word relations is a part of a part of the the title, then it that just doesn't even begin to cover the relationships that you're that you're building and maintaining through your work.

JackIf you were, like, an exec at like a big kind of massive DevTool, would you put metrics like of equivalent size of a cloud in area or something? Would you be putting any kind of metrics or would you sort of measure it differently?

KimI mean, I think you still have to. Right? Like, at the end of the day, even though leadership does understand the value of DevRel a little bit better now, you're still gonna be held accountable. Right? So there's there's always going to be, you know, some tie in of, like, here's the value we're bringing. And if if you can represent that value in, you know, the number of active participants in your beta program or something, and that can be recognized as bringing value to the product that's not necessarily, like, directly financial, then you can you can start to work with some more of the the metrics that before we were just kind of arbitrarily assigning dollars to. But you do have to have like that kind of understanding with leadership, and I think that's that's where things have changed is leadership does have more of an understanding of that.

JackI guess it's kind of an extension of, yeah, just like sales and marketing. There's always had that trade off and fight between them. Right? Whereas like, you know, I I worked in sales back in the day and, you know, like, all leads are not born equally. You know, it was like kind of and you might you know you know, the like the marketing team is measured on like how many leads they send the sales team and then it's like they would want to send like

KimLike your MQLs and your SQLs and

JackYeah. And they would wanna like if you if you think that people optimize what they're measured on, then it's like probably the most rational course of action would be to just send you as many leads as possible that meet the minimum criteria and sort of, you know, whereas from like a absolute optimization of, you know, revenue, they would baby behave differently. And it's like, I guess if for salespeople, it's kinda easy in a way because but there's also other like weird incentives, I guess. But like yeah. It's I guess it's really hard when when you're like you're not the one at the finishing line.

It's kinda hard to to measure it, to have like a metric that is aligned with the business, I suppose.

KimWell, exactly. And, like, that was what the the framework was trying to create a way for executives to to explain, like, the financial value of DevRel to executives in an era where we had to do that or, you know, like, we'd lose our jobs basically, which it's not really seen that way anymore, especially with product. Like DevRel that's in product are very closely related to product because when DevRel is very closely tied to product, they care more about DevRel helping the product experience and helping existing customers than, like, speaking at events or, you know, having a having a booth or or showing a logo or something like that. Like, there's definitely the, like, there's the marketing side and then there's the product side. And both of them have issues.

Right? Like, especially because a lot of times in a lot of orgs, if you're in one or the other, you don't really have a strong tie to the other one. That's, you know, kind of where you get the, like, oh, I got some really valuable feedback, but I'm in marketing, so I'm just gonna throw it over the wall, the product, and then it disappears and I never know what happens. Or if you're in product and and, you know, you're really focused on improving the developer experience, but you don't get to go out and talk to developers, like, then you have kind of the opposite problem. Right?

So I think it's that's also one of the reasons that that, like, I really like working at smaller companies because those lines are much more loosely drawn. It is a lot easier to talk to other areas of the company, but everybody, you know, kind of knows each other.

JackAs you go forward with Keycard, do you think that you all have, like, things that you're gonna start measuring? Like, or or you just keep just focusing on like kind of because I I feel like in my startup DevRel experience, there's been like very much like vibes at the beginning where it's like kind of drawn to the thing that's like the most pressing at that time. It feels very much like that. But we were very much like yeah. We're very

KimI I do think like metrics will become very important because kind of without them you can't demonstrate growth. So even if you aren't necessarily looking at like, okay, what's the equivalent financial value of x y z? If you're not tracking anything, then you don't know if your product is is gaining traction with developers at all. So there's there's definitely, you know, probably going to be a tie like, soon as we start getting, like, a regular content publishing cadence going, things like that, then, you know, we're we're obviously gonna start measuring things. But it will probably be largely focused on kind of like the a flywheel motion because we are early enough where everybody is going to start at the beginning of the flywheel.

Like, we're not working with a situation where, like, oh, we already have a champions program, and then we have people who are dabbling, and then we have enterprise customers, and we have all of these things. Like, being able to start at the beginning and figure out what to measure as it's literally taking shape, I think is gonna be exciting. And I'll probably have, you know, completely different views about all of the measuring frameworks and all the, you know, flywheel stuff that I've done in the past after having, you know, kind of gone through product going to market for the first time.

JackYeah. Could you actually share a little bit about the flywheel for people that haven't come across it?

KimSo basically, it goes in five stages. Okay. So this is this is basically what it looks like. This is an article that I wrote in 2022. And this is basically like this is still how I think about this. Like, this is the thing that hasn't changed. And I think the Keystone stuff, you know, I think about that a little differently, maybe gravitating more strongly in this direction where you have this cycle where there's people who don't know anything about your product. So you want to inspire them to be interested in your product in the first place. And I like kind of this idea more than awareness because just because you're aware of something doesn't mean you're ever going to use it. And so awareness is not really like the greatest thing to focus on, I think, because what you're actually trying to get to is interest.

So in the first section, you know, you're trying to give people information that shows how something would be valuable to them or interesting to them or exciting to use or useful. And then they become, you know, interested in actually going and maybe looking at your docs because maybe they've seen a demo or something. And then you get into kind of the educate phase where people are exploring at like an informational level. Like, maybe this might be a product that my company would use, so I'm gonna I'm gonna read up about it. I'm going to kind of look at what other people have had ex you know, what other people's experiences.

I'm gonna see how the APIs work. I'm gonna read the docs. Or these days, like, I'm gonna ask an LLM. Is this product any good for my use case? And then you kind of get into activation. So once they've learned enough about it to be willing to actually try it, then as soon as they start using it or, like, they download it or they install it, then then they're they actually do become users. And getting to each of these, like, phases basically takes you know, there there's like a jump in between all of them. And the experience has to be good in every single one of them because if you mess up the education experience and you have terrible docs, then people aren't even gonna get to activation. Right? Like, they're gonna be like, well, I have no idea how this works, so I'm gonna just move on.

And when people do that, you often lose them forever. Like, you don't necessarily get those people back. So I think that, like, the these first couple of stages are really important that you get it right. Because by the time people are using something, they're usually a little bit more forgiving because they're kind of committed to trying to get something to work. I mean and that's not to say like the rest of it isn't super important, but like these two phases are where you get the most drop off and developers are just we're unforgiving.

Like, if I have a bad experience, I'm probably not gonna come back to the product. That's it's just like the cold hard truth. But then once you have people, you know, who are using, then how can you actually engage with them in a meaningful way where, you know, you are actively helping them and you're actively able to maybe get information from them that can continue to help improve the product. So you have, like, your engagement phase where they are committed enough to actually, you know, talk to DevRel or have the have a con be willing to have a conversation with, like, a product manager about their experience or something like that. And then once you kind of get through that, then basically you're getting into your champions at that point.

Right? So I there's like these the the people here. Right? Like, so at the top, you know, you have your general users, your developer audience, general developer audience. And then here, you're you're in the builder phase. Like, people educate themselves and then they transition into into actually trying it and then they play around with it. And then they get to be power users, you know, somewhere along here. And then, you know, if if you're lucky and they like you enough, then they become champions, which it that doesn't necessarily mean like they join a champions program or something. It it might mean that they talk about your product, and they tell other people, this is awesome. You should use it.

And at that point, then you wanna empower those people to continue to to so those people inspire more people. So basically, you get kind of that that's how the flywheel goes. And in the very beginning, you know, it's like it's just you. At the very, very beginning, it's just you, the company, doing these things. But once you actually get to a point where you have passionate developers using your product and talking about it or making videos about it or sample apps with it or talking about their enterprise launch with it, then those people are then inspiring other people.

And so the company as a company, you wanna empower those people to have the resources to do what they're doing. And and I think, like, that's where a lot of champion programs kind of come in is giving resources to the people who do care enough to actually tell other people about it and give them tools to be able to move into kind of inspiring other people.

JackThis is so so great, Kim. Yeah. It's yeah. That's it's really, really cool because it helps me to solidify, so you kind of wanna make sure you go for all these steps so that you can get people I mean, firstly, it just creates a great experience and then also can lead to like, because everyone talks about like word-of-mouth and how that's the best way and like you wanna grow by people. You know, that's how like I think most of us discover tools is through our friends and stuff.

And then this is like a

Kimfriendly Social proof. Social proof is huge. If somebody uses something and they're like, no, this really sucks. I'm probably not even gonna try it. You know? And if a bunch of people use something and they're like, this is great, I probably will try it. And, like, that's you know, having that social proof is incredibly important. So, yeah, like, the article on the whole basically just says what I just said, but also gives examples of what happens in each of these phases. So these are things that you could measure these things. Right?

Like you could take this and actually measure components of the examples of what happens within each of those sections. But it is, you know, like, it's a gradient. Like, you don't just jump from one to the other usually. Like, you move along the line until you hit, you know, an inflection point where you are willing to go the next step. And I think, you know, part of the flywheel success from, like, a company standpoint is making it easy to get from step to step.

Like, if you have people who are using the product, but then they don't know how to start engaging with the company, but you don't have a a community or you don't have a way to actually talk to them, then they might be potential champions, but they'll just never get there because you're you're not giving them a way to engage. So stuff like that. Yeah.

JackYeah. That's very cool. So you're gonna you're gonna follow this with Keycard as you go forward?

KimYes. Yeah.

JackThat's awesome. For sure.

KimBut we're, I'm like just trying to get into the inspire cycle right now of like trying to produce content that will inspire people. And and yeah, like then I'll put a bunch of effort into the docs because that's, you know, where I'll start working on educating and and everything. So it'll go I'll be stepping along through it with people from the beginning, which is which is just nice because I don't have people in every stage already that I'm trying to retain or, you know, trying to get over the next hurdle because there's there's there aren't people there yet. Yeah. Just You the green field is just so rare.

JackYeah. Yeah. I can make a terrible joke that you don't have to worry about the whole wheel at the moment. You only need to worry about like a curved arrow. But inspire.

KimYeah. Exactly.

JackYeah. That's awesome.

KimThank heaven because I'm a team of one at the moment. And and I have like four jobs really. So Yeah. Yeah. Wear lots of hats sort of situation.

JackYeah. That's amazing. Well, I guess like one last question. If you have anything that, you know, you would want to really just say to someone at a DevTools start up, like if you took away nothing else, just remember this. Or like something that you really you tell a lot of your friends and people that you come across in DevTools.

KimI think, honestly, right now, it has to be related to AI because there's just so much stuff happening and there's so many people who are also, like, worried about their jobs and worried about the future or really excited about the future. And there's not really a whole lot of people in the middle I have noticed.

JackAnd,

Kimyou know, I've been reached out to by people that I've worked with in the past who are really concerned. Like, they're like, do you think that I'm going to have to switch careers? Because, like, my job will be, you know, done by agents soon. And I think, like, the the takeaway that I would say having come from a place skepticism, like strong skepticism to, you know, where I am now and kind of realizing just how much of a multiplier it can be is that it doesn't using it doesn't make you dumber. It doesn't make you, like, lose your own expertise or your knowledge.

It actually, like, makes you need to be a little bit more precise because you're telling something that kind of things for itself, but kind of also doesn't I do think of AI as like a very young child with a super high IQ. Like, it knows a lot of things, but it doesn't always know how to apply them. And it doesn't always know when it's just making shit up or, you know, when it's kind of using something inappropriately. It's just like, I'm smart. I'm gonna run with it.

And so human expertise, being guardrails around that is critical. Like right now, that's kind of I think the difference between slop and actually building something with AI that works and is relatively bug free and provides a good experience. So I guess like as an example, like when when I was implementing Keycard with, you know, the the app that I was trying to build, you know, agentically, there were several times when I had to say, no, you got this, like, I am concept wrong. Or no. Like, this this is how this works, and you're, you know, reading the wrong documents.

You're referring to the wrong thing. And, like, if I didn't have that expertise, there's no way I would have known that. I would have just let it keep doing whatever it was doing wrong. And there's been tons of times where I've watched it struggle and spin, and then I'm like, why don't you just do this super simple obvious thing? And it's like, oh, you're absolutely right. You know? However so much. But it's it really is like something that needs a human element right now. And I think that the most important way to like kind of be successful and kind of make sure that your your job in tech has longevity is learning how to use it effectively. Because if you are the one creating the tasks or, you know, the goals for the AI, like, then you're you become an orchestrator.

Like, your your job responsibilities might shift. Like, you might not be writing, you know, hundreds of lines of code yourself, but you almost get to like at this higher level where your your decisions are even more important and like even more critical. And I think that that, you know, that was like the advice that I gave to someone who reached out to me about, you know, about being concerned about it. And having used it, I definitely feel even more, like, value in the expertise that I've built up throughout my career. Because now I get to apply it in a way where it really, really matters to everybody who's going to use this thing that I built, like, into the future.

And without without that expertise, it it literally would have built Slop. Like, that's that's the kind of the difference. And then also just, like, identifying where can I build something with AI that will actually help me, help me be more productive or help me save time? And building those tools is something that yeah. Like eventually AI and in some cases, probably AI does build the tools, but the more sloppy put in, the more sloppy you get out.

So it is kind of like, if you tell something that builds kind of crappy things to think of more things to build crappily, then you just kind of get more and more of things that are worse and worse. And we need people to say, oh, this is the human need that we have right now. You know, I'm going to try to architect something that will solve this. And the tactical way I'm going to do that, the execution is that I'm going to use agents. So I think that like people who are worried about just their jobs purely getting eliminated, I would say just try to learn, build something with it or build an MCP server.

Just build something really simple and straightforward that helps you specifically and learn how it works. And I think that, like, the more people who see it working too will realize that how valuable their human experience is at this, like, point in time.

JackYeah. That's very good advice. Kim, thank you so much. Where can people learn more about you and about Keycard?

KimI have my website is maida.kim. And that's where all of these articles about, like, the flywheel and and whatnot are. Keycard, you can go to keycard.ai. We are working on updating the site, but also I'm working on a pretty large blog post that talks about this big project that I've been dogfooding, which I use every single day, multiple times a day, like as a practical tool. And I'm actually pretty excited to to share with people, like, the capabilities that it has and also, you know, how I integrated it with Keycards.

So that will show up on Keycards blog. And, so by the time this comes out, it'll probably be out, so go look for it. And and I am on X as Kim Maida and LinkedIn. And, yeah, my X account was compromised a little while ago, but I did get it back. So it's actually me again now.

JackOkay. Well, you need some secure a in was it insecure agents? Agents weren't secure. I

Kimdon't I mean, maybe. Like, I don't

JackYeah.

KimI don't know. But they they, like, they got paid for premium. So when I recovered my account, I had premium for a little while on, like, somebody else's dollar. That was that was kinda funny.

JackSaw your post on this, actually. Yeah.

KimYeah. You you put on LinkedIn

Jackthat was

KimIt was was murky.

JackAwesome. Well, thank you so much, Kim. Thanks, everyone, for listening.

KimYeah. Thanks for thanks for having me.

JackThanks to Dan for making it happen. Dan Moore.

KimYeah. Yeah. Definitely.

Open the transcript file

Hosts and Guests

Hosts

  • Jack Bridger Host Feed-stated

Guests

  • Elliott Roche Producer Feed-stated

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. Siddhartha Borah from Brackett: teaching agents work that was never written down (Heavybit DevGuild)

    Sep 3, 202611 minAudio

    In this episode, Siddhartha Borah, cofounder and CTO of Brackett AI, joins us at Heavybit DevGuild. Siddhartha explains how Brackett is building agents that can learn complex business work - especially in areas like...

    Credit the people on this episode

  2. James Arthur from ElectricSQL - agents are data, not compute (Heavybit DevGuild)

    Aug 27, 202612 minAudio

    In this episode, James Arthur, cofounder and CEO of ElectricSQL, joins us at Heavybit DevGuild. James explains why Electric has evolved from Postgres sync into data infrastructure for collaborative multi-agent systems....

    Credit the people on this episode

  3. Dave Fletcher from LeadDev: what engineering leaders want from AI

    Aug 20, 202613 minAudio

    In this episode, Dave Fletcher, cofounder of LeadDev, joins us at LDX3 London. Dave shares what LeadDev is seeing from thousands of engineering leaders, including what tools they are buying, how AI coding tools are...

    Credit the people on this episode

  4. Dennis Pilarinos from Unblocked: context for AI coding agents

    Aug 17, 202618 minAudio

    In this episode, Dennis Pilarinos, founder and CEO of Unblocked, joins us at LeadDev in London. Dennis explains why AI coding agents are only as good as the context they can access, and why every new agent is like a...

    Credit the people on this episode

  5. Ian Coe from Tonic.ai: synthetic data, AI training, and enterprise fine-tuning

    Aug 6, 202611 minAudio

    In this episode, Ian Coe, Co-founder and CEO of Tonic.ai, joins us at Heavybit DevGuild. Ian explains how Tonic started by helping developers create safe, realistic data for testing and development, and how that has...

    Credit the people on this episode