150: Everybody Loves Jev  🔗 **Source:** [150: Everybody Loves Jev](https://share.snipd.com/episode/0e14d8f4-3828-43cc-ac96-127e34dd806d) ## Highlights - **Solo Parenting Rewards Improvisation** - Solo parenting four children became easier when the five-year-olds started helping with younger siblings. - Ian improvised a “dad shower,” served burnt fries as “crispy fries,” and learned that children adapt when routines change. Transcript: Ian Landsman Yep it was uh actually better than i had feared all right i love my children dearly it's the logistics of solo parenting that always terrify me um and i think it's getting easier the five-year Can help themselves obviously right but can also start like helping me and helping little kids get ready and it is it's completely different obviously terribly exhausting but for Example uh know, I'm loading them all into the minivan on Sunday morning to get to church. So I got them up. I got them dressed. I got them fed. And I'm buckling my two-year daughter into her car seat in her pretty dress. And she looks down at her legs and starts laughing and says no pants no pants and i had forgotten to put i had forgotten to put shorts on under her dress so i'm like okay but then my one of my Five-year piped up and said i can run upstairs and get them and i was like what are you what you can't go yes do it go inside run upstairs get some shorts um so i think it's getting better yeah Aaron Francis It does get better the big kids helping is a nice unlock for sure even if it's just they're at the phase of like go grab this for me like it's still they can go grab stuff accurately most of The time without most of the time unbuckle everybody bring them inside the whole thing yeah and the other thing is i always like the dynamic i mean your kids are still a little you're a Little bit in the like uh you know survival mode obviously when they're real little but that's why i always encourage jamie to go away because there is a nice element of like the whole Vibe is different when it's you with the kids and you just get a little different vibe you know and it's just like yes this is a nice change i get a little direct one-on time they can't go To mom for anything like it's me and them and it's just like i could do weird dad stuff and that's our weird dad time you know so there's it's nice to occasionally have that time to yourself With the kids. It is. Ian Landsman You know, Jennifer got home on Sunday afternoon while they were asleep for rest time. So the big kids have rest time. The babies have nap time. So, like, everybody gets up at 3 o'clock, 2.45, 3 o'clock after rest time, and immediately it's, mommy, mommy, I'm, you know, hanging on mommy. I'm crying to mommy. Mommy picked me up. And I told Jennifer, uh, I was like, man, it was actually a lot easier before you got home. They just, they're just like, well, mom's not here. I guess I'll just sit quietly in color and, you know, or I'll wrestle with dad or, you know, whatever. So, yeah, it is. And I also think it builds robustness because I'm like, oh, guys, I, you know, I forgot to do this. Sorry, we're just going to have to deal with it. And like, OK, that sounds great. Or like Saturday night, I was like, man, I should give all four kids a bath. I'm not going to do that, though. And so I invented the dad shower where I just got a towel and got it really wet and then wiped down their arms and legs and face and put them in bed. Aaron Francis Oh, boy, breaking the rules there. It sounds like, oh, baby. Yeah, exactly. It can't always be the same, even for them. You know, you got to have some things change. The world changes on you sometimes. You got to be ready for it. ([Time 0:01:22](https://share.snipd.com/snip/963c35ee-e7e9-4799-ba1d-d6925c86b24b)) - **Learn Microeconomics Before Giving Business Advice** - Learn basic microeconomics before confidently discussing how businesses and customers behave. - Aaron recommends asking an AI tutor to create a month-long course covering supply, demand, motivations, and trade-offs. Transcript: Aaron Francis Can the people just take microeconomics? I know you're not going to go to college. Don't go to college. Just go to the Claude or the Chachi PT and literally say, I want to learn the basics of microeconomics. I want you to build me a course. We're going to do it every day for a month and you can get through it. Ian Landsman Based on everything you've seen on X.com, the everything app, do you think there is any possibility that anyone in the world is going to do that? Aaron Francis You know what I'm trying to do here? I'm trying to reach one person. If I can reach one person who's like, you know what? I'm on here pontificating about a bunch of shit I don't understand. I'm going to listen to Ian. I'm going to learn about microeconomics so I have some idea what the hell I'm talking about. Ian Landsman Underneath all your doomerism, you have this unbelievable optimism that somebody is going to go create an economics course for themselves this is the dumbest thing i've ever heard Okay so as an accounting major and economics minor so i i'm studied in the ways of economics yes just quietly and calmly set the scene for me. What are you seeing on Twitter and what do we need to discuss here? I don't know. Aaron Francis I can't even. I don't. Just the whole thing is like. Spiraling. Just give me an example. Just give me an example. Nobody seems to understand why people buy anything. Like this is like a foundational thing that bothers me. They don't understand supply and demand, and they also don't understand the motivations. They don't understand trade-offs. They don't understand... My ergonomics word is skipping out of my mind at this point. You should make a course. I should make a course myself so I can remember the exact term I'm thinking of. It starts with a C, but I can't remember right now. But anyway, there's the reason you need to understand why people do the things they do in order to effectively sell them things. And I think that this is just a foundational thing that people should work on. ([Time 0:09:59](https://share.snipd.com/snip/5fca1c81-30e9-42f7-9e0c-3e3763e3cf34)) - **AI Erodes SaaS Moats Without Erasing SaaS** - AI reduces the effort required to replace simple SaaS, but that does not make replacement economically wise. - Developers may spend days rebuilding tools despite maintenance costs, creating competitive pressure without eliminating the category. Transcript: Aaron Francis Okay, so there was, he's referencing a different tweet, so it's going to be a little complicated. The tweet he references is somebody replaced their mintlify which is just some ai coded thing that like got thrown together a couple weeks ago they're replacing the mintlify with uh You know they they replaced themselves with claude okay took them 12 hours right so their original poster says uh something like my mintlify thing my mintlify renewal was coming up. Ian Landsman It was like obscenely expensive. Mintlefy has raised all their prices. I also think that's the company where the CEO told everybody not to smile, but I digress. So Mintlefy renewal was coming up. The guy was like, you know what? I'm just going to rip it out in one Clodder Codex session and then be done with it. Now, that's the original tweet. John, what's his name? John Youngfook? Aaron Francis Youngfook. Ian Landsman Youngfook quote tweeted it. Give us the John tweet. Aaron Francis There's a whole category of SaaS predicated on the moat of devs don't have time to do this themselves. And that moat is eroding in real time now. And this is just the kind of useless bullshit that is not helping anybody no this is not even a real thing it's not a real thing tell me tell me piece piece by piece so like let's let's let's Ian Landsman Steal man it. What is the truth of what John is saying just from like a good faith reading of it? A good faith reading of it, right? Aaron Francis Is there, I mean, is there any tools in the whole world that devs didn't have time before to build themselves and now they can to replace? Yes, of course, there are probably some. Ian Landsman You're having a bad faith tone while saying good faith things. So just balance that out, but keep going. Aaron Francis Should that be a thing you actually consider? I would say for 95% of like bootstrappery type people or just anybody selling software, that's not that important an angle on. Okay. Ian Landsman So if I may speak like claude for a moment yes you've found the load-bearing seam and honestly it matters okay so what you said is should it does matter the thing that you said matters a Lot so the thing that you said was is there some category of sass that could be replaced with ripping out tokens. Yes. Should you consider doing it? Absolutely not. I agree with all of that. Okay. All of that. So like we have a firm foundation upon which we agree. Okay. The second layer is, so the original statement was, should you spend your time and tokens doing that? No. Meaning, should you just buy the product that solves your problem? Yes. Is that a fair, like, therefore it follows? Okay. Now the important thing, and it matters, is how, in all of history, how many times have developers done things that are a poor ROI or poor use of their time? They do that all the time. So I think if we're to continue steel manning, I think what John could be saying is it is not wise. He didn't say this, but we'll put these words in his mouth. It is not wise for you to take three days even to ViCode a Mintlify replacement. It is not wise. That could be inferred. He didn't say it, but it could be inferred. That's not what he's saying, but okay. No, no, no. But if we're, so this is the Aaron saying, if I'm John, I'm saying, is it wise for you to do this? Probably not. Are developers going to do this 100% Okay. And so I think what John is saying is things like Mintlify, things like Banner Bear, which is his company, part of the moat is it would be hard for a developer pre-AI to spend a month or two Or three on it, which they shouldn't. But now they can spend two, three, four days a week on it, which they shouldn't. But they are willing to do that. How does that, how does that land? So what I'm separating is your advice is to the bootstrapper that says, and the advice is don't freaking do this. My commentary is the reality that says bootstrappers are going to freaking do this. ([Time 0:12:47](https://share.snipd.com/snip/8bdb933e-c3d2-41fc-b76f-646d8763fb9e)) - **Companies Buy Software To Avoid Permanent Maintenance** - Company build-versus-buy decisions depend on maintenance, continuity, and organizational incentives—not merely whether a developer can create a prototype. - A developer may vibe-code a replacement, but companies usually avoid owning another system forever. Transcript: Aaron Francis No. Okay. So this leads in perfectly to why I think people should take microeconomics. Sure. Because what he's saying is this category is going away. Okay. And that's fine for an individual dev might make this individual decision. So if you sell a product that's for individuals, that's for individual developers, then yes, that could be problematic. I'm totally on board with that. That's possible. I don't think that's who Mintlify is trying to sell to. I don't think it's for individual developers. And I don't think most software is for individual developers. And I don't think most software developers are making decisions at all for themselves because they work in some big company that's making decisions as a company. And in that scenario, I do not buy the premise that like software developers inside companies are going to spend even a fair amount of time replacing other tools with homegrown hacky Things that then they will have to manage forever, even after this person leaves in two years, because they're leaving in two years. So now the team has to manage their own Slack because they built their own Slack and it doesn't make any sense. So this is why the micro economics comes in, because why do companies build things? Why do companies buy things? What do companies build? What do companies buy? What are the tradeoffs? Like, whereas so much of Twitter is like, me developer, I can build a replacement to Slack today. I can vibe code it in one $200 token thing. It's going to do all the stuff we use Slack for today, right? I can get rid of it. That's true. But nobody's going to do that. ([Time 0:17:12](https://share.snipd.com/snip/89da7037-b940-467b-98ac-bcf904531371)) - **Behavioral Economics Explains Vibe-Coded Replacements** - Behavioral economics explains why companies may act against rational cost calculations. - Ian points to Cursor replacing Sanity and companies building narrow internal substitutes, even when buying would be wiser. Transcript: Ian Landsman Oh yeah, he's dropping the trends. Part of economics, there's a branch called behavioral economics. Part of like pure economics relies on reasonable actors. The rational person when seeing a $5 bill on the ground would pick it up. The economist when the economist when seeing a $5 bill on the ground would not pick it up because they would say someone will have already picked that up. Right. And so there's like a difference between what is like pure economics and what is behavioral economics. I think what you're describing relies on rational actors. Like should companies, I don't think Slack is a good example because I think almost 0% of people believe that they could take on Slack. I think Mentlify is a great example because honestly it's docs hosting. And so I think you're saying rationally should a company devote any time to docs hosting themselves? Approaching 100% no, if I may give it a Jev probability here. Approaching 100% no. But will they is a very different question. I think it's an extremely different question. It is. We see it already. No, we don't. This is my point. No, Cursor. We don't see it already. Cursor, owned by Grok, owned by SpaceX AI, Elon Musk's holding company. Right. Cursor replaced Sanity.io with their own internal vibe-coded slop fest. Which, is it a good idea? I think you and I would probably say no. Like, go train some new models. What the hell are you doing? Did they do it? Yes, they did. And we're seeing, you know, the plural of anecdote is not data, but we are seeing many anecdotes of people actually in companies, not just like indie hackers in the Philippines, actually In companies saying, I need 80% of what this SaaS, no, I'm sorry, I need 20% of what this SaaS offers. I'm going to vibe code a replacement for just my 20%. Like that is, that is actually happening. Now. Now, is John's point that like the whole industry is going away? No, I don't think. I think that's overblown. But people are actually doing this. ([Time 0:19:05](https://share.snipd.com/snip/0ce5f73e-c8ad-4ff3-9eba-54dcdeac77a5)) - **Aaron Builds An AI-Planned Makerspace** - Aaron leased a 1,000-square-foot brick commercial space to combine an office, studio, and makerspace. - He used an iPhone scan and AI-assisted Blender modeling to plan furniture and equipment before moving in. Transcript: Ian Landsman I signed a lease for a commercial space. I signed it to start October 1st. Then they did like some fire sprinklers and Electra. And so it's going to be November 1st, unfortunately. I went over to the space. So this is, first of all, this is paid for by Aaron Francis Holdings Incorporated. Just so any Laravel Holdings Incorporated employees that are listening are like, what the hell? Why did he get an office? The new guy got a commercial office and I'm stuck in my basement? Okay, this is coming from whatever's left over from Try Hard Studios Treasury. Yes. So it is a commercial space, about 1,000 square feet, brick, internal brick on like three sides, concrete floors, exposed joists, high ceilings. It's beautiful. Beautiful. Yeah, this is so much better than the other place you were talking about. Aaron Francis I'm so happy you end up in this place. The strip mall, yeah. The strip mall is going to be really grim. You're not going to feel grim going to this place. You're going to be very happy. Ian Landsman Massive skylight in the middle. Oh my gosh. Building of other people. I am definitely going to do a party. Architects, event planners, like all kinds of creative spaces, which is super fun. So i went over there sometime last week um for the express purposes of 3d scanning the place so i found this app in the app store called polycam that uses just a standard iphone to do a 3d Scan of a space and so walked around the whole space like four times up down left sideways uh and i got this 3d scan and then i took you know probably a hundred pictures just like of every Little thing turning in a circle and then i took a few pictures with a tape measure present so i could get some like ground truth realities of like actual dimensions. So then I come back here to my office. I load every I get everything off my phone, put it in a folder and tell Astra, good luck, have fun, use Blender to recreate this. And then I hop on, you know, a million calls for work. Right. So when the day's over, I pull up Blender, or no, I'm sorry, I tell Astra, make me a webpage where I can walk through this. I don't want to open Blender, in fact. So I pull up the 3D model in my browser, and I start walking around inside the space, and it was incredible. So I had to do a few, I had to do a few like, hey, this, you know, this landing is too far out or like the handrail doesn't actually land on the concrete here. I need you to adjust that. So I don't want to claim it was one shot because it definitely wasn't. It was probably eight shot, which is not bad. So just going back in and being like, this part feels wonky, check the 3D, check the images, make it feel correct. And so it took the 3D scan and it put it in the viewer as well, but the cooler part is when you turn off the 3D scan, you get the very clean blender render. Because the 3d scan is all that like weird kind of like mesh stuff and this is this is like perfect geometry if you just turn that off and so now i've started i've started basically taking Measurements of like the chalkboard and the couch that i have and other stuff. And now I've said, all right, let's make some new 3D assets. And we're going to plop some furniture in here. And you're going to help me like, you're going to help me set up my studio based on the actual stuff I have and the actual space I'm going to be in. And it is so cool and so valuable. ([Time 0:33:35](https://share.snipd.com/snip/66b0c6f2-d957-443a-bce8-5139759884c6)) - **Jev Turns Fuzzy Decisions Into Cheap Classification** - Jev is optimized for fast, inexpensive structured classification rather than generating prose. - Product builders can use it for fuzzy branching decisions, replacing expensive model calls or brittle hand-coded heuristics. Transcript: Ian Landsman JEV is a new System 1 model, which I have never heard this term before, System 1 and System 2. It's a new thing. JEV came out to much fanfare last week. And the premise behind Jev is LLMs are overpowered for classification or decision tasks, right? So you ask Codex, even soul high. We won't even go to Astro. Yes, Codex soul high. Does this look like spam? It's going to take four seconds and it's going to give you back a a written uh answer of yeah it looks like spam because of xyz and it's going to cost we'll say two cents right the theory behind Jev is it actually cannot produce text at all so it's not an llm it cannot produce text it can only produce structured responses. And therefore, because of some training mechanism, whatever, it is like a hundred times cheaper to run than one of these frontier models. So you would say, here's an email. Here's my question. Here are, here's the set of valid choices, which could be spam, help, billing, whatever. Or you could ask it yes, no questions, and it'll give you a confidence. Or you could ask it to score something, which if yes, no gives you a confidence and a score is afloat as well, I don't understand why those are two separate concepts. But that is the idea behind JEV. And so from my perspective, JEV is not useful whatsoever in a coding sense. But for product builders who are trying to build in workflows, switch statements, branching logic, and all of or some of those decisions can be somewhat fuzzy you can replace your hand-coded Heuristics or potentially expensive llm call with a very very cheap heuristic ([Time 0:50:01](https://share.snipd.com/snip/4539ee2b-5813-443b-9fa4-e0d02a0ca188)) - **Jev Cuts Labeling Costs With Escalation** - Ian classified 94,000 terminal snapshots with Jev for about $9, then planned to escalate disagreements to frontier models. - Jev achieved roughly 70% accuracy, making it useful as a first pass rather than a final authority. Transcript: Ian Landsman So I think treating Jev as a smart heuristic is the move moving forward. I don't think you should treat Jev as an oracle that will always give you the right answer. I think you shouldn't treat any LLM that way. But if you can do a first pass, let's take Solo, for example. Soloterm.com, it's very good. I'm building this classifier engine, right? So over the weekend, I let Jev rip on 94,000 stored terminal grids. So in my development, not on production, in my development, I capture these terminal grids every 30 seconds just so I can get training data. So as I'm going about my normal work, I'm getting all this training data. I now have 94,000 of those terminal grids. And what I have done historically is grade every single grid using both Codex and Claude. And if they both agree that a grid is, you know, asking for permissions or working or whatever, then we count that as successfully labeled. What I did over the weekend is I graded all of them with Jev, all 94,000, and it cost me $9, 227 million tokens, and it cost me $9, which is awesome. And so what I have done now is I've just knocked out half of my cost for frontier model labeling. So what I'll do now is I'll have one of the frontier models go through and label some subset of that, not all 94,000. And if it agrees with JEV, then I count that as past. If it doesn't agree with JEV, then we adjudicate to another frontier lab. A permissions prompt and Jeb says this isn't, we'll escalate to Claude and say, what do you think? And then we'll go with the agreement there. So I think that is a responsible way to slash costs and maintain quality by having it. Every grid is still double labeled. One is just by a classifier. One is by a frontier model. Aaron Francis Not bad. Did you have it? What was its accuracy compared to the double LLM check? Ian Landsman To the double LLM check on a small sample set, it was like 70% accurate. Okay. So it's not as good as two Frontier models, which is fine. I'm okay with that. Aaron Francis For your use case, that seems reasonable. You're still going to have a check with a frontier model. So you're still actually, so now you're probably only a few percentage points below the accuracy level you were at. Although I don't know. I don't, I mean, I don't know enough about the math of your particular use case where like maybe that is important. Maybe like being two or 3% less accurate is actually going to make people quite annoyed. How am I two or three percent less accurate? Because presumably the two frontier models acting together are going to classify something correctly 99 percent of the time. Yeah. And now you're only going to be now with Jev and one frontier model, they're only going to be accurate 94 percent of the time. Ian Landsman Because you think when Codex is wrong, Jev might also be wrong, and then Claude could have saved it, but it'll go through if I do the agreement. Aaron Francis Now you don't have the two strongest models, right? You have one much weaker model and then one heavier model. So presumably there'll be some that get through that wouldn't have gotten through. There has to be some, or otherwise Jev's as good as Claude, which isn't true. So there has to be some that make through now that wouldn't make through before. ([Time 0:55:36](https://share.snipd.com/snip/a677b856-5ccb-4527-93ed-658da3869661)) - **Use Cheap Classifiers Where Errors Are Tolerable** - Use Jev for broad checks such as spam, newsletters, and out-of-office messages, while reserving stronger models for consequential decisions. - Evaluate error costs in context; a few percentage points can meaningfully increase unwanted alerts. Transcript: Ian Landsman How do you plan on using Jev inside of Help Spot? Aaron Francis So, I mean, the thing that we've taken a look at quickly is stuff like a spam check stuff like there's a pre flight process for like, is this a type of email that should does this contain Basically like a question? Or is this like a newsletter? Or is this spam? Or is this something that like, we shouldn't spend an out of office reply? Is it an out of office? Yes, like all those kind of things. So that's where like Matt here has taken a look at Jev and like it seemed reasonable for those kind of like broad stroke checks. But we'll see. It's not going in anything to start. It might eventually make its way in. To me, that's where like the backup for JEV, if JEV is ultimately the best one of these, I mean, who knows? Ian Landsman Like I imagine within a month or two, we're going to have the OpenAI and Anthropic version of this, right? It's a bloodbath. I would never compete in this space. Aaron Francis Right. So we'll see. Whichever one ends up being the best will be the one we use. And then to me, the backup, well, whatever. I guess this is different. If we have three good ones, then the backup will just be the other good ones, Like if they're all equal. Ian Landsman If we have three good ones and they're all this cheap, run them all. Aaron Francis Run them all at the same time. Right now today, I probably wouldn't use Kev as my backup. I'd probably use Luna as my backup. It's like, okay, like when this fails, I'm just going to go to Luna. That's what I'm doing now. So like I'll just fall back to that and it'll be slower and it'll be like a tiny bit more expensive. Again, people get caught up in the expense, but in practice, like it doesn't really matter. It was below a cent and now it's one cent. It's pretty irrelevant for like what I'm doing. Sure. I'd rather save the half a cent, but I also kind of don't care. So again, like again, if you're putting it in your switch statement or whatever, inside your code. If it's a billion half cents, I can't do that math. Ian Landsman But if it's half a cent times a billion, that's a lot. Aaron Francis So that'll be interesting, right? Are there businesses that are possible? Are there tools that are possible because you can call something with some logic a billion times in some sort of reasonable way? Probably, right? So that'll be cool to see what's possible there. So I was very excited as soon as it came out, but then everybody ruined it. Everybody makes it terrible. Nobody ruined it. With all their stupid takes. Ian Landsman Nobody ruined it. Those are discrete variables. They're uncorrelated variables. Jev is cool. People on Twitter do stuff for hype. Uncorrelated. Aaron Francis Here's what I think for the people listening to this show, my takeaway would be to really think about it and how you can do something useful with it. That's not the information you're getting. 95% of the information is stupid and you should ignore all the stupid information and think about actual practical uses to solve actual business problems and that'll lead ([Time 1:00:07](https://share.snipd.com/snip/4a762ce0-ccd8-4f39-b3bc-22386135b8bc)) - **Raise Prices When Your Product Creates More Value** - Raise prices when inflation and new AI-driven efficiencies provide credible justification and market cover. - Even a 20% increase can outperform modest churn, especially when the product saves customers money or expands their revenue. Transcript: Ian Landsman Should we be charging more or less for software oh we gotta talk this is this is ian being mad episode so jesse handley by the way friend and uh financial patron of the show. So, you know, tread lightly. Jesse had a tweet about, you know, this is basically the worst time to raise your prices. You should be lowering them. Ian comes in over the top with a quote tweet, slam dunk and says, 100% wrong. No freaking way. Raise your prices. Laura Roeder weighs in. All right. What's the story here? Aaron Francis I didn't technically say 100% wrong, but I do think that it is a good time to raise prices because you have a lot of cover from inflation. Everybody's raising their prices. QuickBooks has raised their prices twice this year. Salesforce just raised their prices. Microsoft just raised their prices. Like lots of companies and software companies are raising their price. Mintlify apparently just raised their prices, right? Like, and we talked about this on the show before. It's just a math problem. And so many like, so many sort of bootstrapper type companies are scared to raise their prices on their existing customers, but you have to do that. It's like just free money. And okay, if you raise your price 20% and 3% of your customers go out and vibe code their own replacement, you are still up a shit ton of money. So that's fine. It's fine if that happens. So, and if you raise your price 20% and you lose 50% of your customers, I think there's something else going on there. Because very few industries are the customers that price sensitive. Obviously exceptions, developers are annoyingly one of them, right? Like maybe they'll do that. Fine. If you sell purely to developers, okay. Like there's probably a few other, if you're consumers, if you sell a to-do app to pure consumers, it's harder to raise your price. Although I've gotten plenty of price increases in the last year or so from the few consumer apps I do subscribe to on my phone. But whatever. More price sensitive for sure. I don't even think that's price sensitive. That tends to be they forgot they even were subscribed to you and you woke them up. And so you just lost 15% of people instantly who were like, well, I didn't even know I was still using fuzzy to do. And like, so I'm just going to unsubscribe. Ian Landsman That's a good name. But anyway, yeah. Aaron Francis So I don't know. I just think that there's a lot of cover for that. First of all, second of all, if you're integrating AI and you're bringing efficiencies, this is why software charges more. It's literally the reason, right? It's like, Hey, if you implement our software, you are going to save X dollars is one of the main marketing strategies in software. And so you're going to save more or you're going to make more are the two main ones. So, hey, you're going to buy our tool and you are going to make more money for your business because you implement our tool. So that's, again, a good reason to charge more. Like I added all these AI features. I'm charging more for them because it is going to save slash make you 10 times what you're paying me, right? So it becomes the no brainer. Of course, I'm going to buy this thing because I get to fire half my support people or I get to sell twice as effectively ([Time 1:04:58](https://share.snipd.com/snip/6dde2597-2b9e-4257-9d6d-61a2d0b648e1)) - **Your Customer Segment Determines Pricing Power** - Pricing advice differs sharply between developer-focused products and software sold to established businesses. - Buyers weigh reliability, deliverability, and maintenance alongside price, which lets products like email infrastructure charge for trust. Transcript: Ian Landsman Think we've found i think we've found the root of all issues okay let's go the root of all issues is 99 of developers on twitter sell to other developers. Right? Including John Youngfook, including Jesse Hanley. Aaron Francis But is that even true of Jesse Hanley? I don't think that's true. Ian Landsman I'm not sure that's true. Let's hold it out as true. Okay, let's just say it's true. Ian, on the other hand, is on Twitter swimming in that water, but not being a native species fish in that water. That's true. Right? So you are in that water, but you are, it's ships in the night, right? All the developers are giving advice to other developers, which maybe on the whole is bad, considering your point of view, which is sell to real companies, sell to institutions, do Not sell to other developers. And so I think two things can be true. In your world, raising prices because of inflation and AI efficiencies makes 100% sense. In Jesse's world, I think who he's speaking to is mostly developers doing developer tools for other developers. And he might be right that now is not a good time to raise prices because hopefully your company is getting more efficient and there are lots of competitors in that space. And there's lots of maybe developers who value their time at negative $100 who are willing to vibe code a replacement. So it does feel like a little bit of a ships in the night situation where you're talking to these people and Jesse's talking to these people and everyone is assuming we're talking to the Same people, but in all reality, we're not. Right. Aaron Francis Yeah, and I think, I don't think that's really true of Jesse, though, if we could talk about Jesse in particular, because, like, he... Okay, let's just say he's only selling to developers, which I think is not true, but let's just say that's true. What he does is very difficult and actually incredibly hard to vibe code to replacement to, right? Like, standing up your own email server and having it be deliverable. No, but that's fine. He doesn't care. Again, like somebody paying Jesse Hanley $5 a month, who's like, oh man, he made it six. I'm out of here. I'm vibe coding my own email sending system. Okay. If you lose that guy, it doesn't matter. Meanwhile, the person paying Jesse $300 a month or $1,000 a month, right, is no way going to make that decision so lightly because they are sending to thousands of their customers. They probably already understand that deliverability is quite hard and they have no reason to believe that they are going to be better at deliverability than Jesse is. And putting aside even all the rest of it, you have to stand up servers and deal with IP addresses, all the stuff that goes into sending an email, right? So like, I just don't think they're that price sensitive. And I think Jesse could probably raise prices still going to be undercutting all the competitors, right? Which I mean, price is always a factor, but it's not the only factor. That's the thing. It's not the only factor. So there are a lot of other factors besides price involved. ([Time 1:08:19](https://share.snipd.com/snip/bd97b080-4c1e-4175-b223-a11fb96bb90b))