# 131: Crush My Competitors  ## Episode metadata - Episode title: 131: Crush My Competitors - Show: Mostly Technical - Owner / Host: Ian Landsman and Aaron Francis - Episode publish date: 2026-05-12 - Episode AI description: They discuss major Solo product changes and unifying internal tools. Hiring stories include a first full-time operations hire and frustrations with resumes. Fundraising choices and tiny-seed tradeoffs come up. Team migration to Codex/Claude and token, logging, and pooling headaches are explored. They end with a light feel-good AI project built for fun. - Duration: 01:06:53 - Episode URL: [Open in Snipd](https://share.snipd.com/episode/88933d48-cc5e-4209-873a-1da62c73447c) - Show URL: [Open in Snipd](https://share.snipd.com/show/7ad45fb4-8aa6-49c2-83cc-bc2327d14b03) - Export date: 2026-06-28T13:38:20 ## Snips ### [Unifying Solo's Operations Layer](https://share.snipd.com/snip/d3052b52-cf86-41dd-987a-a6b934176647) 🎧 05:44 - 11:44 (05:59) <iframe src="https://share.snipd.com/embed/obsidian-player/snip/44a92b11-83b2-4bea-b5e9-bf1f5c288565" width="100%" height="100" style="border: none; border-radius: 12px;" sandbox="allow-scripts allow-same-origin allow-forms allow-popups allow-clipboard-write" ></iframe> - Aaron is consolidating Solo's internal ops so HTTP, CLI, and the MCP share a single operations layer to enable multiple ingress points. - This refactor unlocks first-party work-tree solutions and lets third parties build custom workflows without diverging code paths. #### 💬 Quote > So I'm taking all the MCP functions and moving them down a layer so that it's just like this generic operations layer. > — Aaron Francis Aaron Francis on Solo architecture refactor #### 📚 Transcript **Aaron Francis:** So I'm trying to make hay because the sun is shining. And I don't know how long the sun will be shining. Right? Okay. And I mean that partially in two regards. One is how long will I have the will of the people with regard to Solo? And how long will tokens be this cheap um because you think about it you're like oh man six hundred dollars a month and you're like you couldn't you couldn't hire you couldn't hire somebody you couldn't hire a remote worker anywhere in the world to do this, like, 24-7 programming for $600 a month. Right. The money is nothing. **Ian Landsman:** No, **Aaron Francis:** it's nothing. Yeah, so last night I fired up another email address that I already have and subscribed and turned it on extra high fast and let her rip overnight. So we're just fast. **Ian Landsman:** I don't like the fast. I don't know. It's not same model. I don't like it. I **Aaron Francis:** think it is. No, it's not. **Ian Landsman:** It's like running on a whole different platform. It's like a different model. I **Aaron Francis:** think it's the same thing. It just costs twice as much. **Ian Landsman:** Yeah, I don't know. The Claude, that's what Claude is just faster. Same model. Code different model. Now, I don't know if they train it similarly. I think they're attempting to get them close, but they're not the same model. **Aaron Francis:** But anyway. I have to look at that. Yeah, beyond the fast. I'm ripping. So I'm in a pretty good spot, I think, with Solo. I've been doing a lot of internal work. I haven't rewritten the terminal again. So I think we're still at version 4.5 or whatever. But I'm doing a lot of internal work to set up my next big moves so I can crush my competitors underfoot, of course. And a lot of that is just like, a lot of that's pretty mechanical. So it's just like moving this stuff around, preparing for that sort of thing. And I'm also doing the Wilking redesign is about to go out and so a lot of a lot of tweaking and tinkering there um but the big internal stuff i'm very excited about because what i'm doing is so the mcp is very robust very strong lots of great uh functionality there the HTTP API was originally built just to like, basically, it was just built for Raycast. So you could like list your projects, you know, start, stop, open, whatever. I've since, and I never published the Raycast extension because I wasn't ever really happy with it. And also nobody's ever asked for it. **Ian Landsman:** Yeah seems like i would ditch that yeah so i i **Aaron Francis:** don't think i'm going to publish it because nobody cares about it um which is great i love not doing things um so but what i am doing is i'm trying to bring the uh http the mcp and the forthcoming CLI all into compatibility. So I'm taking all the MCP functions and moving them down a layer so that it's just like this generic operations layer. And then each ingress point will have its own little thin shims. anything you can do with the MCP, you can do with HTTP, you can do with the CLI. And that is something that people actually are asking for. What are they asking for? The CLI? They're asking for CLI and HTTP because they want to build solo into their workflows in a more like automated way. And once this is done, it also opens up what is going to be my solution for work trees. So I have in my head my solution for work trees and I'm not going to say it here because I don't want my enemies to to scoop i think **Ian Landsman:** you're overthinking this **Aaron Francis:** but okay no no i'm under thinking it so i don't i don't want my enemies to scoop me but once i get all of this done then i can do i can do uh i can implement my solution for work trees or on write clones or discrete checkouts or discrete clones or basically any other thing that you want to build there. Okay. And this gives the people, so I'll ship first party solutions to that, but it also gives the people the universal escape hatch to get exactly what they want. Because everybody's like, I want a work tree, but I don't like how Conductor does it. But I don't like how they do it. But I don't like how this comes. And I'm like, OK, well, what do you want? And like, oh, I just want something that's this super specific thing that's totally weird and makes no sense. And I'm like, OK, well, I can't ship that. So what do I do? But I have the solution and it's very good. It's very good. Yes. I have overthought the solution and it's really good. That's why we need a third Codex **Ian Landsman:** account. I got you. **Aaron Francis:** That's why we need a third Codex. So we're getting there. So I think big, big things, big things are coming on the horizon. We've got the redesign, which is looking very good. We've got windows, which is almost there. And then we've got this unification project. And those three will ship probably in the next two weeks, I think. --- ### [Grab Users First Then Build Features](https://share.snipd.com/snip/fbc1f1e9-1e78-4b2c-ac6f-d0b3d57a4764) 🎧 13:33 - 16:10 (02:36) <iframe src="https://share.snipd.com/embed/obsidian-player/snip/f77be8ec-11e9-4866-be08-60a0202877c6" width="100%" height="100" style="border: none; border-radius: 12px;" sandbox="allow-scripts allow-same-origin allow-forms allow-popups allow-clipboard-write" ></iframe> - Prioritise land grab growth over building deeply esoteric features during early adoption phases. - Ian urges shipping Windows support and videos to capture users now, then add advanced features over the next years. #### 💬 Quote > We're in land grab mode. > — Ian Landsman Ian Landsman on product focus during early adoption #### 📚 Transcript **Aaron Francis:** Because **Ian Landsman:** here's the thing. We are in land grab. No, you haven't got me. I got you. haven't me at all. He's speechless. A few videos that are unedited and not out yet do not have me. So here's the thing. We're in land grab mode, okay? Land grab mode. There will be time in the future to build all the other features. That time is in the future. That's not now. Now it's time to get people using it, and then you add the features behind them. Time **Aaron Francis:** is a construct, so telling you we have user numbers are growing i will say good i like they're growing money money is being made not hand over fist uh but money money is being made users are signing up um it is shocking to me i mean it's i use solo all day every day but to hear people say like solo is the tool that i spend most of my time in i'm like whoa that's crazy um so people are using it we're getting more people in the door uh yes i'm trying to push forward on all fronts simultaneously which turns out is difficult. That's **Ian Landsman:** what I'm saying. I feel like the thing to me is like once you're on your third codex, I feel like you're in too much time coding. Too much time. I'm **Aaron Francis:** not even coding. Codex is coding. Too much time **Ian Landsman:** thinking about coding, analyzing code, thinking about features, coming up with new stuff too much time too much time and listen this is the same experience i had with help spot which is there was a land grab everybody's like oh let's get rid of this this uh client server help desk software we're all going browser-based right and that's the most i ever sold was like in terms of not dollars but in terms of new users right is that first like three years and because that's when the big push that's when everybody switched. And then after that, it's so much harder to get customers after that. Like it's just, there's so much more competition, blah, blah. And this is the same phase, right? Like everybody's going to lock into what they use for AI. They're locking into their models. They're locking into their tools, blah, blah, blah. You want to get as many of those beautiful butterflies as you can right now. You got to grab them. And then you have 10 years to add the features behind. I **Aaron Francis:** don't know if that's right. You do. There is an initial bar that has to be overcome. Sure. But you already asked that. People **Ian Landsman:** are using it. They like it. They're using it all day, right? --- ### [Aaron Moves To A Studio And Hires First W-2](https://share.snipd.com/snip/63215659-22b8-4fed-8493-ae301581ec69) 🎧 22:31 - 29:58 (07:27) <iframe src="https://share.snipd.com/embed/obsidian-player/snip/52f899d3-c1c8-45d9-a220-f2bfaaaf3f84" width="100%" height="100" style="border: none; border-radius: 12px;" sandbox="allow-scripts allow-same-origin allow-forms allow-popups allow-clipboard-write" ></iframe> - Aaron is moving Try Hard Studios into a commercial 20x50 space and hiring his first full-time W-2 employee to run operations. - The new director of operations will handle sponsorships, bookkeeping, taxes, and scaling day-to-day business tasks. #### 💬 Quote > This new employee is brought to you by Bento... Hiring my first full-time employee whoo. > — Aaron Francis Aaron Francis announcing a commercial space and full-time hire #### 📚 Transcript **Aaron Francis:** Uh, and I think I've mentioned it here on the show. Um, I'm moving out of the apartment. **Ian Landsman:** Uh, yes, we have talked about that in the show, but it's, it's a done deal. We are, we **Aaron Francis:** are moving soon. All the inspections and everything have passed. Um, and so now I just need to sign the lease. So I found a space. So for the dear listener, I work out of an apartment. I do not live in an apartment. I live in a house with a million children. And so working from home for me, looks like drive three minutes down the road, come into this apartment that is an apartment. And this is the bedroom that I'm in. And database school is recorded in the living room out there. And it's awesome. You know, got private bath, private kitchen, got a pool the kids can use, got a gym that I never use. It's awesome. It's a great scam. however i am one constantly living in fear of a fire inspection and the fire chief walking in and saying what's behind that wall and me answering the window that your firefighters would come in should something happen and him saying uh well you can't build walls in front of windows. They teach us that in Fire School 101. So that's one thing I live in fear of. The other thing is I just, I'm constricted by the layout of the space. These rooms are made for living, not for recording. And so, you know, this one's 10 by 10 and that one's maybe 11 by 11. And so anyway, I really should have a commercial space. And I found one. So I found one like five minutes down the road. It's 20 by 50 and it's just a blank white rectangle. No walls, no fixtures, just there's a bathroom at the back and it's just a white rectangle. So I will be moving out. It's a storefront, yeah. So I'll be moving out over the summer into that space. And this is maybe breaking news, hiring my first full-time employee. But! Hiring a W-2. Wow, that's a big step. Big step. **Ian Landsman:** Are you going to use like a JustWorks or how are you to manage the whole? We already have Gusto. We have Gusto **Aaron Francis:** for that. Does do that stuff? **Ian Landsman:** It does like the payroll? Yeah. Or is just bookkeeping? Okay. **Aaron Francis:** No, it does the payroll. That's how I pay myself. So, yeah, Kelsey is moving on to greener pastures. She's going to a farm up north. She's not dying. She's just, she's got a big gig. She's got a big gig. So I knew all along that Kelsey would not last forever because she's just incredibly smart and talented. But her life situation worked out with my work situation where she didn't want full time and she just wanted to be at home and work part time. And I was like, oh, this is perfect. Well, the day has come where she's going to move on to this incredible, I think, full time gig that I won't talk her her book on, but it's way better than anything I could ever offer her. her and she's excited and sad but she's gonna um help me transition to this new employee uh so she we still work together but she's just able to do fewer and fewer hours as uh the transition starts um and so this new employee is someone that i've worked before we get to the new employee i feel like we should pay our own bills here so we my goodness. What a professional. This new employee is brought to you by Bento. The email, marketing, and CRM platform built for the AI era. Send your product marketing and transactional email with Bento. You can find them at bentonow.com and brought to you by Native PHP ultra. Now that native PHP for mobile is free, get access to all first party premium plugins, AI skills, and priority support for you and your team with native PHP ultra from $35 a month. And friends of the show, use code mostly technical for $100 off the yearly plan. You can find them at nativephp.com/ultra. And finally, this employee is brought to you by Vask. Vask is pusher compatible web sockets, but build sanely. No fan-out fees. It's powered by Cloudflare. Get their first three months free on any tier using coupon code Mostly or Technical. So you have an option there. I'll be curious. A experiment. You can find Vask at Vaskk v-a dot deb so thank you to bento native php and vask for sponsoring this show and my first full-time employee whoo **Ian Landsman:** so what's this person going to do kind of what kelsey was doing on steroids you got other stuff planned kelsey on steroids is exactly. So **Aaron Francis:** worked with this person. I worked with her for maybe, I want to say five years. And she was she was like my right hand, my right hand person for a long time and ruthlessly competent, just like brutally competent. And so she is going to be, I don't, I probably won't give her the title like president because, you know, you got to have like title growth capabilities, you know. So she's going to be, she's going to run the business. That's going to be her whole job. You run the business. She's the director of operations. Yeah. And I have told her your job is to make us both a **Ian Landsman:** lot of money. That's your job. She's going to listening to this podcast. She's going to be on my side of these debates, by the way. No, no, no, no. Ian has a point. Ian has **Aaron Francis:** a point. Fiduciary duty to be on my side. So I've told her that in her purview is obviously, you know, sponsorships, contracts, deals, all of that kind of stuff, but also bookkeeping, taxes, talk to the accountant, find us a business credit card that I can spend money on and get points and cash back and stuff. The whole Aaron empire is under her control. That is her job is to make sure that in my bank account, I have more money, which means I have more money to pay her. So that's her entire job. My job is to create a spectacle. Her job is to run the business. So I am so excited, 100% excited to have somebody full time and somebody local. --- ### [Applicants Are Playing The Wrong Game](https://share.snipd.com/snip/4ffea163-6014-41b1-87ca-3da89e8cbf51) 🎧 31:54 - 34:56 (03:02) <iframe src="https://share.snipd.com/embed/obsidian-player/snip/1240f61d-6818-47a4-9330-885ff9038fbd" width="100%" height="100" style="border: none; border-radius: 12px;" sandbox="allow-scripts allow-same-origin allow-forms allow-popups allow-clipboard-write" ></iframe> - Job applicants treat hiring like a game of luck by mass-applying rather than optimizing one strong tailored application. - Ian and Aaron note resumes with basic errors and irrelevant bullets signal lack of skill and attention to detail. #### 💬 Quote > They have mistaken the game that they are playing. > — Aaron Francis Aaron Francis on poor resume strategies #### 📚 Transcript **Ian Landsman:** I don't even understand what these people are doing. I don't understand what goes through your head when you're like, let me apply this job and let me put no effort at all. Let me not, I mean, just this last week when I was raging on Twitter about it, it's forget the stuff if you customize it for the job. Forget all that. Just straight garbage generic resumes. Just garbage. Just errors all over the place. Identical paragraphs describing two different jobs. Dates backwards. applying for a job that's a hands-on contributor job. But every single bullet point in your entire resume is about managing people and productivity increase amongst the managed people. It's like, what are we doing? What are we doing? I **Aaron Francis:** can tell you what we're doing. Oh, bad. I can tell you exactly what's happening. They have mistaken the game that they are playing. So they are playing a game of luck. They think, rather, they're playing a game of luck. Exactly. And the best strategy for a game of luck where is free, so it's a game of luck where you don't have to pay to play, is to play as often as you possibly can. That's true. Right? That's social media, right? Yeah. It's a game of luck. Great things could happen on Twitter, and the best way to play that game is to just tweet a lot and see what happens. It's a free ticket and you might win a bunch of money. Contrast that with a game of skill. The ideal strategy for a game of skill is to put a lot of work into one specific entrance into the pool. And so they think they're playing a game of luck, which is I'll send out a thousand resumes and maybe I'll get lucky and somebody will call me back. But they're really playing or should be playing as a game of skill where I make sure that my entrance into this game is highly weighted and I put a lot of effort into it. And then that one entry will get me a win versus a thousand entries that might **Ian Landsman:** get me a win. They're just playing the wrong game. This is, you're exactly correct. And it's completely analogous to poker where people think they're playing a game of luck and so they don't study and they don't learn and they never win, right? Because they think they think they're going to get lucky and win but they don't get lucky and win because they're just bad and they lose generally speaking almost every time whereas if they put in the skill and they got a little skill in the game they could actually win most of the time even though it's actually there is an element of luck to the game of course and just like this there is an element of luck like do i read your resume? Don't I? Do I happen to see it? Am I a bad mood? Blah, blah, blah. There's some luck in the job application process. But there's also skill. --- ### [Tailor One Strong Job Application](https://share.snipd.com/snip/3c50cdac-be4d-478f-b6dd-7d8be50dc564) 🎧 34:10 - 35:44 (01:34) <iframe src="https://share.snipd.com/embed/obsidian-player/snip/16971364-9f15-4737-ad2b-9463a6de58a8" width="100%" height="100" style="border: none; border-radius: 12px;" sandbox="allow-scripts allow-same-origin allow-forms allow-popups allow-clipboard-write" ></iframe> - Do treat each job application as a game of skill: craft one high-quality tailored submission rather than spraying generic resumes. - Proofread, correct dates, and mention relevant skills like AI to show competence. #### 💬 Quote > The ideal strategy for a game of skill is to put a lot of work into one specific entrance into the pool. > — Aaron Francis Aaron Francis giving job-application strategy advice #### 📚 Transcript **Ian Landsman:** get me a win. They're just playing the wrong game. This is, you're exactly correct. And it's completely analogous to poker where people think they're playing a game of luck and so they don't study and they don't learn and they never win, right? Because they think they think they're going to get lucky and win but they don't get lucky and win because they're just bad and they lose generally speaking almost every time whereas if they put in the skill and they got a little skill in the game they could actually win most of the time even though it's actually there is an element of luck to the game of course and just like this there is an element of luck like do i read your resume? Don't I? Do I happen to see it? Am I a bad mood? Blah, blah, blah. There's some luck in the job application process. But there's also skill. You have to send in something that's not shit. Like, that has **Aaron Francis:** to be the case. Spray and pray is a great strategy for a game of luck. If there are a thousand games going on and you can put one ticket into each game for free, that's awesome. But if there's one game, if there's one game going on, you have to put in some effort if you hope to win. So you didn't find anybody. **Ian Landsman:** Well, ultimately, was i think a couple people that are worth interviewing so we're gonna see there was a couple but i mean i'm talking about two out of like you know and this is already pre-filtered so i had already gone through and pre-filtered so i wasn't even looking at the like absolute garbage so this is like 60 or whatever resumes of like --- ### [AI Team Plans Need Token And Control Polishing](https://share.snipd.com/snip/445e0a4a-7726-4bae-8763-9acc074bcc1c) 🎧 43:29 - 48:16 (04:46) <iframe src="https://share.snipd.com/embed/obsidian-player/snip/d5e472d2-3d86-4adf-a575-8f7c69370e1f" width="100%" height="100" style="border: none; border-radius: 12px;" sandbox="allow-scripts allow-same-origin allow-forms allow-popups allow-clipboard-write" ></iframe> - Team AI subscriptions currently lack ergonomic controls like token pooling and clear rate vs weekly limits, creating friction for teams. - Ian moved the team to Claude/Codex and found rate-limit increases misleading and token logging/options immature. #### 💬 Quote > If I'm paying for three seats as a team, why are we not pooling tokens? > — Ian Landsman Ian Landsman on shortcomings of team AI subscriptions #### 📚 Transcript **Ian Landsman:** So this last week, I also moved everybody on the team over to Claude and Codex team plans. How's **Aaron Francis:** that going? That's freaking **Ian Landsman:** terrible. It's a nightmare. It's the worst experience in world. It's so expensive. I'm not even into, like, I've only crossed over into actually paying for extra tokens a couple times and i'm at like 10 extra dollars so i'm even saving money but it's like the freedom of like i don't have to think about the tokens is just gone on this and it's because they won't let you buy the 200 a month plan flawed let you buy 100 but even that seems kind of like nerfed a little bit. Although they did come out at the of the week when they did this whole XAI. They acquired a bunch of capacity from XAI. And so they doubled the usage limits and stuff. So that was probably better. **Aaron Francis:** Kind of. We'll talk about that in a second. Yeah, we can talk about it **Ian Landsman:** in second. But I don't know. It's just like you need the team accounts to do this properly because like there's just you need the team controls like everybody just on their own like this doesn't work like we had to let a guy go there a month ago i had to just like ask him to like turn off his thing like there's no i need to be able to like turn things off but like whatever and i don't know there's also this whole other weird thing so there's that side of it like the very businessy side of it. Like people, you hire people, you fire people. We need to be able to manage that. There's like, but then there's also just things like, I don't know. Like they're letting, like part of it with Impetus was like they let, Claude built a security tool that I would like to use. I thought it was for teams and enterprise it's only for enterprise so i still can't use it even though i'm on the teams so hopefully that'll roll out to teams soon um but they are they do seem to be releasing stuff to the teams at enterprise before the public on certain features not on everything so it's like that's kind of interesting. Yeah, I don't know. The other thing that's sort of weird is like, and I do feel like I don't like this, but it's like it has all the logging stuff, like enterprise logging stuff. Like I don't even want my own stuff logged. Like I just want to turn all that shit off. But you can't turn it off unless you're on the enterprise. And it's like, I want to be on the enterprise because I just pay per token, which is way too expensive. So it's just a lot. I feel like they're doing all this stuff, shipping all these features. Yep. I need them to stop. Yeah. The same discussion I've given you. No, no, no. It's different than me, but yeah, I agree. them stop and I need them to like double back and like clean up some of what they've left behind them. I And just lock things in a little bit. Because I feel like it's a little scatterbrained right now on both Claude and Codex's team plans. And we need a little refinement phase there. **Aaron Francis:** Slow down for thee, but not for me. I agree with you for them. I disagree with you for me, obviously. Yeah, I agree. I think the, I have to imagine the logging is useful i know some companies have come out and said like hey listen we're paying for all your tokens those cannot be used for side projects and so i imagine that's that's helpful there i don't know i haven't i haven't seen what the logging even looks like so I don't know if it's helpful there. So to circle back on the five-hour thing, such a scam. It's not, it is, they didn't increase, like, how many tokens you can use per week. They just said the five-hour rate limits are increased, so you can just burn your weekly plan faster. And it got all this, like, it got all this applause for, like, oh, they increased their usage limits. They didn't. They just increased the rate limits. **Ian Landsman:** Yeah, but see, that's really important. That's actually the most important part to me. Because there's a lot of times where I don't use the whole week very often of limit. But I do a chunk like when i moved over to teams i was instantly like i did like nothing and i was like 20 of the five hour window and the problem is once you cross the five hour window you have a choice right like you can set it up to just stop working which i think isn't ideal or you can set it up to go into token using paid tokens and so that's where i'm going to use a lot of my paid tokens. It's like I have this block. Today's the day I can actually just code. I'm going to blow through the five-hour window, and now I'm going to be paying for tokens, even though I'm going to, in the end of the week, not have even used my full lot of tokens. All right, that is frustrating. So that part is super frustrating. I even think there should be like, I think all the tokens should be pooled. That's another thing. If I'm paying for three seats as a team, why are we not pooling tokens? Because not everybody's using it the same amount. Right? **Aaron Francis:** Great point. Yeah. --- ### [Security Work Delays Token Town Launch](https://share.snipd.com/snip/169fbbb7-37f0-44fa-87f5-7d37239c5b46) 🎧 53:57 - 56:22 (02:24) <iframe src="https://share.snipd.com/embed/obsidian-player/snip/1aee95b0-6e29-4afb-a35c-8489876ef342" width="100%" height="100" style="border: none; border-radius: 12px;" sandbox="allow-scripts allow-same-origin allow-forms allow-popups allow-clipboard-write" ></iframe> - Ian paused Token Town launch to focus on HelpSpot security and SOC2 work because AI makes automated exploit discovery easier. - He prioritized security hardening and logging to prevent automated AI-driven multi-step exploits. #### 💬 Quote > The AI is going to get really good at like not just finding an issue, but being like, oh, I can string 12 of these things together into an actual exploit. > — Ian Landsman Ian Landsman explaining security focus and delays #### 📚 Transcript **Aaron Francis:** The world is evolving. I didn't think that was what it was **Ian Landsman:** going to be. have to stop what we're doing and we have to do some heavy security stuff on help spot on surrounding supporting things for help spot because it's just getting too easy for anybody to be like hey ai like go in here and like hack me this thing and we at least have to have some semblance of you know we've done various sorts of security stuff over the years, but not in this AI era, the full deep dive. So as I talked on the show, built like AI security tooling, trying to use the Claude security thing if they ever let me in. I have used the Codex as a security thing, which I am in and have used, although the way it works isn't my favorite because it kind of like goes off PRs, which are commits, which is fine, but can only go back so far. So I have like an app that's older than going back so far. So it just won't look. It doesn't go back 20 years. Yeah, it won't go back. So like helping old code get checked, at least not directly. So blah, blah. So that's just it. So we've spent a lot of time on that. We spent a lot of time on getting, we kicked off our SOC 2 efforts to make Userscape SOC 2 compliant. Sounds miserable, but probably important. That's very But also important, also along the lines of the security stuff. So it's kind of like security vibe. Couple weeks, couple ones here. But that stuff's rolling. Plugged a lot of bugs. Fixed a lot of things. A couple real ones, a couple like on its own is probably not the hugest deal, but like this is one of the biggest fears about the AI era and security, right? It's like the AI is going to get really good at like not just finding an issue, but being like, oh, I can string 12 of these things together into an actual exploit right and so like so you kind of really are forced to fix a lot of these things that maybe in the past you would have pushed or maybe nobody would have found because it's like a weird internal escaping thing that we'd have to have like an ex privilege escalation first which requires all those other steps to even get to that point. But it's like, well, now like that's a thing. Like let's assume they're going to make it seven steps in. Well, I know you should always have assumed that the whole time, but whatever. We were living in the real world, but that was a hard thing to do. And unless you were selling a product that China was going to come try to hack you, probably you could survive without it. --- ### [Daughter Builds Horse Name Generator With AI](https://share.snipd.com/snip/8472774b-ecd1-432c-9d2d-cbb665e0ab18) 🎧 59:35 - 01:05:47 (06:12) <iframe src="https://share.snipd.com/embed/obsidian-player/snip/428e22d7-0240-4558-857d-a6d39c04dd4b" width="100%" height="100" style="border: none; border-radius: 12px;" sandbox="allow-scripts allow-same-origin allow-forms allow-popups allow-clipboard-write" ></iframe> - Ian and his 13-year-old daughter built a horse-name generator site using Claude, then published it on GitHub at MiaLandsman.com. - The project engaged his daughter, taught publishing basics, and produced a shareable app with favorites stored in localStorage. #### 💬 Quote > We have a horse named Jenna she was so excited she's like this is the greatest. > — Ian Landsman Ian Landsman describing building a horse name generator with his daughter #### 📚 Transcript **Ian Landsman:** ai story for you i **Aaron Francis:** love a feel good ai story last week **Ian Landsman:** me and my daughter we're having a little daddy daughter time she came downstairs talking about something whatever blah blah so i talk about horses ball walks into the horses as we all know update on the horse we still don't have a horse but of course you can follow up with that later **Aaron Francis:** but at least you're paying a ton of money no **Ian Landsman:** i'm not paying any money right now i'm not paying any money that's good i'm i'm only paying for lessons and stuff which is like normal money so yes but we will be paying a lot of money soon but we aren't yet anyway i was like hey let's like build an app to help us with this horse naming situation so we had a claw we use clawed and we had to build up like do the artifact so it could like show a html page and we're in clawed and we're going through it and then she was loving it she takes over the laptop she's typing it improving it a lot like now we got this big horse name generator that knows all the types of horses and all the stuff and then we went we went all the way with it we were like all right we're gonna put this on github and explain a little bit of that and then we're gonna like we already had her name registered so if go to mia landsman.com can see this website so we got the name pointed at the github the hosting and the thing and whatever and uh yeah so now there it is we have a horse named jenna she was so excited she's like this is the greatest um i love this she's sharing it with her friends all the stuff it was very cool uh and it was great because none of my kids like coding so this was a nice little. So **Aaron Francis:** this is your 13-year Mia? **Ian Landsman:** Yeah. This is the youngest. This is adorable. **Aaron Francis:** Yeah. Isn't it cute? Crimson Lady, Soaring Atlas, Twilight Willow, Jolly Apple. These are great. These are great horse names. Says how, you know, their **Ian Landsman:** breed and like how much they, how high they jump and all the stuff. And, you know, some different, depending on certain settings, like other things happen. So there's a little bit of that stuff going on. It was really cool. So I recommend everybody do that with the old kiddos when they get big enough. There's way to track your favorites. **Aaron Francis:** You can click the heart button and it tracks the favorites. Puts **Ian Landsman:** it in the local storage. So it's a cool little ai name generator thing i i feel like these guys should probably make a way that's a little easier to publish what you build um i agree that's not really in there right now so you still have to kind of know all this other stuff to do it but uh yeah it was really cool little horse name generator if you have a kid into horse names horses check out MiaLandsman.com. **Aaron Francis:** Well, this is really, this is feel good. This is really adorable. Yeah, it's so cute. This is not dissimilar, like obviously the tech is, but this is not dissimilar from how I got into coding. It was like, I want to do a thing. Let's build a little HTML page and then we'll throw it up. And then, you know, you go to school and you're like, I have a website. And the kids are like, you have a website? What do you mean you have a website? So presumably she's at, you know, she's at lunch and she's going to MiaLandsman.com and she's shown her friends that she built a website. This is great. I love this. Yeah, it's **Ian Landsman:** so good, right? This is pure. It's so pure. And yeah, that's the other thing. It kind of also is making me feel like, you know, people are worried about how you're going to have software developers and stuff. But it's like how people are going to get into it. But I feel like this is just like we used to have to do view source on a website we like and figure it out. This is just like a nicer version of that. Instead of viewing source on somebody else's website, I could build my own website. And they could do stuff that I find enjoyable. But now, like, if I don't have me sitting there, right, she would have to figure out, well, how do I publish this on a domain or whatever? And then that kicks you off on, like, how does that stuff work, right? And you just go down that same road we've all gone down to figure out how stuff works. But, you know, your entry's a little different. You don't have to struggle so hard to actually figure out how it works because the tool can tell you how it works. So, yeah, I thought that was pretty interesting. But, yeah, she got a big kick out of Doing stuff with your kid like that is fun. **Aaron Francis:** Yeah. How was it being dad and getting to show her some visibility into what you do all day? Yes. This is a **Ian Landsman:** big problem I have. It's like the kids don't understand what you do. No idea. You're just on the screen. Like you're looking at the screen. What does that mean? Money comes in. We go do stuff. But the money's all virtual, right? Like, you know, they don't see money. They don't see the effort other than dad looks You don't come home **Aaron Francis:** covered in coal dust. Like, they have no idea what you're doing. And, like, they **Ian Landsman:** look at a screen all day too, right? Like, they're on the screens for school, for when they're home. And **Aaron Francis:** you're like, get off your screen, but I gotta go on the screen. Right. And it's like, oh, so they **Ian Landsman:** look at the screen, I look at the screen, that must be the same thing. So it's very interesting to like just give them just a little bit of insight into like how this works, what I do, even though it's not obviously direct, but it's like the website, how we create a website, blah, blah, what you might do with that. So that part's really cool. Definitely with the oldest, well, all three kids. The oldest, we did some HTML by hand stuff way back in the day. The middle kid, we like a, I actually did like a, what was like learn the code for kid things over the summer? Me and him did that. We were trying to build a game, which I actually think is a terrible starting point. I don't know. Everybody wants to have people build games. If you're really into it, if they're self-motivated to do that, think that's fine. But I think just starting there is a high bar. So that was fine, but didn't go super great. Never got a moment like this. This was by far the best moment of all of them in a coding task. So it was really cool. The AI facilitating the human interaction. Not all bad. Not **Aaron Francis:** all bad. Yep. So it was cool. So go to MiaLandsman.com to check out Mia's horse name generator. Subhead where every horse finds the perfect name. It's great. This is so adorable. I love it. Yeah, it's great. Tell tell me that your weird internet friend thinks it's awesome i **Ian Landsman:** will do and uh yeah that's the part that's cool too it's like it's actually doing something right it's not even just like oh it's my website and there's a picture of me up there it's like yeah oh it's like it's an actual application that actually does something that these kids think about a lot and stuff like that. --- Created with [Snipd](https://www.snipd.com) | Highlight & Take Notes from Podcasts # Snips updated after 2026-04-05T10:16:17 ### [Unifying Solo's Operations Layer](https://share.snipd.com/snip/d3052b52-cf86-41dd-987a-a6b934176647) 🎧 05:44 - 11:44 (05:59) <iframe src="https://share.snipd.com/embed/obsidian-player/snip/44a92b11-83b2-4bea-b5e9-bf1f5c288565" width="100%" height="100" style="border: none; border-radius: 12px;" sandbox="allow-scripts allow-same-origin allow-forms allow-popups allow-clipboard-write" ></iframe> - Aaron is consolidating Solo's internal ops so HTTP, CLI, and the MCP share a single operations layer to enable multiple ingress points. - This refactor unlocks first-party work-tree solutions and lets third parties build custom workflows without diverging code paths. #### 💬 Quote > So I'm taking all the MCP functions and moving them down a layer so that it's just like this generic operations layer. > — Aaron Francis Aaron Francis on Solo architecture refactor #### 📚 Transcript **Aaron Francis:** So I'm trying to make hay because the sun is shining. And I don't know how long the sun will be shining. Right? Okay. And I mean that partially in two regards. One is how long will I have the will of the people with regard to Solo? And how long will tokens be this cheap um because you think about it you're like oh man six hundred dollars a month and you're like you couldn't you couldn't hire you couldn't hire somebody you couldn't hire a remote worker anywhere in the world to do this, like, 24-7 programming for $600 a month. Right. The money is nothing. **Ian Landsman:** No, **Aaron Francis:** it's nothing. Yeah, so last night I fired up another email address that I already have and subscribed and turned it on extra high fast and let her rip overnight. So we're just fast. **Ian Landsman:** I don't like the fast. I don't know. It's not same model. I don't like it. I **Aaron Francis:** think it is. No, it's not. **Ian Landsman:** It's like running on a whole different platform. It's like a different model. I **Aaron Francis:** think it's the same thing. It just costs twice as much. **Ian Landsman:** Yeah, I don't know. The Claude, that's what Claude is just faster. Same model. Code different model. Now, I don't know if they train it similarly. I think they're attempting to get them close, but they're not the same model. **Aaron Francis:** But anyway. I have to look at that. Yeah, beyond the fast. I'm ripping. So I'm in a pretty good spot, I think, with Solo. I've been doing a lot of internal work. I haven't rewritten the terminal again. So I think we're still at version 4.5 or whatever. But I'm doing a lot of internal work to set up my next big moves so I can crush my competitors underfoot, of course. And a lot of that is just like, a lot of that's pretty mechanical. So it's just like moving this stuff around, preparing for that sort of thing. And I'm also doing the Wilking redesign is about to go out and so a lot of a lot of tweaking and tinkering there um but the big internal stuff i'm very excited about because what i'm doing is so the mcp is very robust very strong lots of great uh functionality there the HTTP API was originally built just to like, basically, it was just built for Raycast. So you could like list your projects, you know, start, stop, open, whatever. I've since, and I never published the Raycast extension because I wasn't ever really happy with it. And also nobody's ever asked for it. **Ian Landsman:** Yeah seems like i would ditch that yeah so i i **Aaron Francis:** don't think i'm going to publish it because nobody cares about it um which is great i love not doing things um so but what i am doing is i'm trying to bring the uh http the mcp and the forthcoming CLI all into compatibility. So I'm taking all the MCP functions and moving them down a layer so that it's just like this generic operations layer. And then each ingress point will have its own little thin shims. So you can do with the MCP, you can do with HTTP, you can do with the CLI. And that is something that people actually are asking for. What are they asking for? The CLI? They're asking for CLI and HTTP because they want to build solo into their workflows in a more like automated way. And once this is done, it also opens up what is going to be my solution for work trees. So I have in my head my solution for work trees and I'm not going to say it here because I don't want my enemies to to scoop i think **Ian Landsman:** you're overthinking this **Aaron Francis:** but okay no no i'm under thinking it so i don't i don't want my enemies to scoop me but once i get all of this done then i can do i can do uh i can implement my solution for work trees or on write clones or discrete checkouts or discrete clones or basically any other thing that you want to build there. Okay. And this gives the people, so I'll ship first party solutions to that, but it also gives the people the universal escape hatch to get exactly what they want. Because everybody's like, I want a work tree, but I don't like how Conductor does it. But I don't like how they do it. But I don't like how this comes. And I'm like, OK, well, what do you want? And like, oh, I just want something that's this super specific thing that's totally weird and makes no sense. And I'm like, OK, well, I can't ship that. So what do I do? But I have the solution and it's very good. It's very good. Yes. I have overthought the solution and it's really good. That's why we need a third Codex **Ian Landsman:** account. I got you. **Aaron Francis:** That's why we need a third Codex. So we're getting there. So I think big, big things, big things are coming on the horizon. We've got the redesign, which is looking very good. We've got windows, which is almost there. And then we've got this unification project. And those three will ship probably in the next two weeks, I think. --- ### [Grab Users First Then Build Features](https://share.snipd.com/snip/fbc1f1e9-1e78-4b2c-ac6f-d0b3d57a4764) 🎧 13:33 - 16:10 (02:36) <iframe src="https://share.snipd.com/embed/obsidian-player/snip/f77be8ec-11e9-4866-be08-60a0202877c6" width="100%" height="100" style="border: none; border-radius: 12px;" sandbox="allow-scripts allow-same-origin allow-forms allow-popups allow-clipboard-write" ></iframe> - Prioritise land grab growth over building deeply esoteric features during early adoption phases. - Ian urges shipping Windows support and videos to capture users now, then add advanced features over the next years. #### 💬 Quote > We're in land grab mode. > — Ian Landsman Ian Landsman on product focus during early adoption #### 📚 Transcript **Aaron Francis:** Because **Ian Landsman:** here's the thing. We are in land grab. No, you haven't got me. I got you. You me at all. He's speechless. A few videos that are unedited and not out yet do not have me. So here's the thing. We're in land grab mode, okay? Land grab mode. There will be time in the future to build all the other features. That time is in the future. That's not now. Now it's time to get people using it, and then you add the features behind them. Time **Aaron Francis:** is a construct, so telling you we have user numbers are growing i will say good i like they're growing money money is being made not hand over fist uh but money money is being made users are signing up um it is shocking to me i mean it's i use solo all day every day but to hear people say like solo is the tool that i spend most of my time in i'm like whoa that's crazy um so people are using it we're getting more people in the door uh yes i'm trying to push forward on all fronts simultaneously which turns out is difficult. That's **Ian Landsman:** what I'm saying. I feel like the thing to me is like once you're on your third codex, I feel like you're in too much time coding. Too much time. I'm **Aaron Francis:** not even coding. Codex is coding. Too much time **Ian Landsman:** thinking about coding, analyzing code, thinking about features, coming up with new stuff too much time too much time and listen this is the same experience i had with help spot which is there was a land grab everybody's like oh let's get rid of this this uh client server help desk software we're all going browser-based right and that's the most i ever sold was like in terms of not dollars but in terms of new users right is that first like three years and because that's when the big push that's when everybody switched. And then after that, it's so much harder to get customers after that. Like it's just, there's so much more competition, blah, blah. And this is the same phase, right? Like everybody's going to lock into what they use for AI. They're locking into their models. They're locking into their tools, blah, blah, blah. You want to get as many of those beautiful butterflies as you can right now. You got to grab them. And then you have 10 years to add the features behind. I **Aaron Francis:** don't know if that's right. You do. There is an initial bar that has to be overcome. Sure. But you already asked that. People **Ian Landsman:** are using it. They like it. They're using it all day, right? --- ### [Aaron Moves To A Studio And Hires First W-2](https://share.snipd.com/snip/63215659-22b8-4fed-8493-ae301581ec69) 🎧 22:31 - 29:58 (07:27) <iframe src="https://share.snipd.com/embed/obsidian-player/snip/52f899d3-c1c8-45d9-a220-f2bfaaaf3f84" width="100%" height="100" style="border: none; border-radius: 12px;" sandbox="allow-scripts allow-same-origin allow-forms allow-popups allow-clipboard-write" ></iframe> - Aaron is moving Try Hard Studios into a commercial 20x50 space and hiring his first full-time W-2 employee to run operations. - The new director of operations will handle sponsorships, bookkeeping, taxes, and scaling day-to-day business tasks. #### 💬 Quote > This new employee is brought to you by Bento... Hiring my first full-time employee whoo. > — Aaron Francis Aaron Francis announcing a commercial space and full-time hire #### 📚 Transcript **Aaron Francis:** Uh, and I think I've mentioned it here on the show. Um, I'm moving out of the apartment. **Ian Landsman:** Uh, yes, we have talked about that in the show, but it's, it's a done deal. We are, we **Aaron Francis:** are moving soon. All the inspections and everything have passed. Um, and so now I just need to sign the lease. So I found a space. So for the dear listener, I work out of an apartment. I do not live in an apartment. I live in a house with a million children. And so working from home for me, it like drive three minutes down the road, come into this apartment that is an apartment. And this is the bedroom that I'm in. And database school is recorded in the living room out there. And it's awesome. You know, got private bath, private kitchen, got a pool the kids can use, got a gym that I never use. It's awesome. It's a great scam. however i am one constantly living in fear of a fire inspection and the fire chief walking in and saying what's behind that wall and me answering the window that your firefighters would come in should something happen and him saying uh well you can't build walls in front of windows. They teach us that in Fire School 101. So that's one thing I live in fear of. The other thing is I just, I'm constricted by the layout of the space. These rooms are made for living, not for recording. And so, you know, this one's 10 by 10 and that one's maybe 11 by 11. And so anyway, I really should have a commercial space. And I found one. So I found one like five minutes down the road. It's 20 by 50 and it's just a blank white rectangle. No walls, no fixtures, just there's a bathroom at the back and it's just a white rectangle. So I will be moving out. It's a storefront, yeah. So I'll be moving out over the summer into that space. And this is maybe breaking news, hiring my first full-time employee. But! Hiring a W-2. Wow, that's a big step. Big step. **Ian Landsman:** Are you going to use like a JustWorks or how are you to manage the whole? We already have Gusto. We have Gusto **Aaron Francis:** for that. Does do that stuff? **Ian Landsman:** It does like the payroll? Yeah. Or is just bookkeeping? Okay. **Aaron Francis:** No, it does the payroll. That's how I pay myself. So, yeah, Kelsey is moving on to greener pastures. She's going to a farm up north. She's not dying. She's just, she's got a big gig. She's got a big gig. So I knew all along that Kelsey would not last forever because she's just incredibly smart and talented. But her life situation worked out with my work situation where she didn't want full time and she just wanted to be at home and work part time. And I was like, oh, this is perfect. Well, the day has come where she's going to move on to this incredible, I think, full time gig that I won't talk her her book on, but it's way better than anything I could ever offer her. her and she's excited and sad but she's gonna um help me transition to this new employee uh so she we still work together but she's just able to do fewer and fewer hours as uh the transition starts um and so this new employee is someone that i've worked before we get to the new employee i feel like we should pay our own bills here so we my goodness. What a professional. This new employee is brought to you by Bento. The email, marketing, and CRM platform built for the AI era. Send your product marketing and transactional email with Bento. You can find them at bentonow.com and brought to you by Native PHP ultra. Now that native PHP for mobile is free, get access to all first party premium plugins, AI skills, and priority support for you and your team with native PHP ultra from $35 a month. And friends of the show, use code mostly technical for $100 off the yearly plan. You can find them at nativephp.com/ultra. And finally, this employee is brought to you by Vask. Vask is pusher compatible web sockets, but build sanely. No fan-out fees. It's powered by Cloudflare. Get their first three months free on any tier using coupon code Mostly or Technical. So you have an option there. I'll be curious. psychological experiment. You can find Vask at Vaskk v-a dot deb so thank you to bento native php and vask for sponsoring this show and my first full-time employee whoo **Ian Landsman:** so what's this person going to do kind of what kelsey was doing on steroids you got other stuff planned kelsey on steroids is exactly. I've **Aaron Francis:** worked with this person. I worked with her for maybe, I want to say five years. And she was she was like my right hand, my right hand person for a long time and ruthlessly competent, just like brutally competent. And so she is going to be, I don't, I probably won't give her the title like president because, you know, you got to have like title growth capabilities, you know. So she's going to be, she's going to run the business. That's going to be her whole job. You run the business. She's the director of operations. Yeah. And I have told her your job is to make us both a **Ian Landsman:** lot of money. That's your job. She's going to listening to this podcast. She's going to be on my side of these debates, by the way. No, no, no, no. Ian has a point. Ian has **Aaron Francis:** a point. Fiduciary duty to be on my side. So I've told her that in her purview is obviously, you know, sponsorships, contracts, deals, all of that kind of stuff, but also bookkeeping, taxes, talk to the accountant, find us a business credit card that I can spend money on and get points and cash back and stuff. The whole Aaron empire is under her control. That is her job is to make sure that in my bank account, I have more money, which means I have more money to pay her. So that's her entire job. My job is to create a spectacle. Her job is to run the business. So I am so excited, 100% excited to have somebody full time and somebody local. --- ### [Applicants Are Playing The Wrong Game](https://share.snipd.com/snip/4ffea163-6014-41b1-87ca-3da89e8cbf51) 🎧 31:54 - 34:56 (03:02) <iframe src="https://share.snipd.com/embed/obsidian-player/snip/1240f61d-6818-47a4-9330-885ff9038fbd" width="100%" height="100" style="border: none; border-radius: 12px;" sandbox="allow-scripts allow-same-origin allow-forms allow-popups allow-clipboard-write" ></iframe> - Job applicants treat hiring like a game of luck by mass-applying rather than optimizing one strong tailored application. - Ian and Aaron note resumes with basic errors and irrelevant bullets signal lack of skill and attention to detail. #### 💬 Quote > They have mistaken the game that they are playing. > — Aaron Francis Aaron Francis on poor resume strategies #### 📚 Transcript **Ian Landsman:** I don't even understand what these people are doing. I don't understand what goes through your head when you're like, let me apply this job and let me put no effort at all. Let me not, I mean, just this last week when I was raging on Twitter about it, it's forget the stuff if you customize it for the job. Forget all that. Just straight garbage generic resumes. Just garbage. Just errors all over the place. Identical paragraphs describing two different jobs. Dates backwards. applying for a job that's a hands-on contributor job. But every single bullet point in your entire resume is about managing people and productivity increase amongst the managed people. It's like, what are we doing? What are we doing? I **Aaron Francis:** can tell you what we're doing. Oh, bad. I can tell you exactly what's happening. They have mistaken the game that they are playing. So they are playing a game of luck. They think, rather, they're playing a game of luck. Exactly. And the best strategy for a game of luck entrance is free, so it's a game of luck where you don't have to pay to play, is to play as often as you possibly can. That's true. Right? That's social media, right? Yeah. It's a game of luck. Great things could happen on Twitter, and the best way to play that game is to just tweet a lot and see what happens. It's a free ticket and you might win a bunch of money. Contrast that with a game of skill. The ideal strategy for a game of skill is to put a lot of work into one specific entrance into the pool. And so they think they're playing a game of luck, which is I'll send out a thousand resumes and maybe I'll get lucky and somebody will call me back. But they're really playing or should be playing as a game of skill where I make sure that my entrance into this game is highly weighted and I put a lot of effort into it. And then that one entry will get me a win versus a thousand entries that might **Ian Landsman:** get me a win. They're just playing the wrong game. This is, you're exactly correct. And it's completely analogous to poker where people think they're playing a game of luck and so they don't study and they don't learn and they never win, right? Because they think they think they're going to get lucky and win but they don't get lucky and win because they're just bad and they lose generally speaking almost every time whereas if they put in the skill and they got a little skill in the game they could actually win most of the time even though it's actually there is an element of luck to the game of course and just like this there is an element of luck like do i read your resume? Don't I? Do I happen to see it? Am in a bad mood? Blah, blah, blah. There's some luck in the job application process. But there's also skill. --- ### [Tailor One Strong Job Application](https://share.snipd.com/snip/3c50cdac-be4d-478f-b6dd-7d8be50dc564) 🎧 34:10 - 35:44 (01:34) <iframe src="https://share.snipd.com/embed/obsidian-player/snip/16971364-9f15-4737-ad2b-9463a6de58a8" width="100%" height="100" style="border: none; border-radius: 12px;" sandbox="allow-scripts allow-same-origin allow-forms allow-popups allow-clipboard-write" ></iframe> - Do treat each job application as a game of skill: craft one high-quality tailored submission rather than spraying generic resumes. - Proofread, correct dates, and mention relevant skills like AI to show competence. #### 💬 Quote > The ideal strategy for a game of skill is to put a lot of work into one specific entrance into the pool. > — Aaron Francis Aaron Francis giving job-application strategy advice #### 📚 Transcript **Ian Landsman:** get me a win. They're just playing the wrong game. This is, you're exactly correct. And it's completely analogous to poker where people think they're playing a game of luck and so they don't study and they don't learn and they never win, right? Because they think they think they're going to get lucky and win but they don't get lucky and win because they're just bad and they lose generally speaking almost every time whereas if they put in the skill and they got a little skill in the game they could actually win most of the time even though it's actually there is an element of luck to the game of course and just like this there is an element of luck like do i read your resume? Don't I? Do I happen to see it? Am I a bad mood? Blah, blah, blah. There's some luck in the job application process. But there's also skill. You have to send in something that's not shit. Like, that has **Aaron Francis:** to be the case. Spray and pray is a great strategy for a game of luck. If there are a thousand games going on and you can put one ticket into each game for free, that's awesome. But if there's one game, if there's one game going on, you have to put in some effort if you hope to win. So you didn't find anybody. **Ian Landsman:** Well, ultimately, was i think a couple people that are worth interviewing so we're gonna see there was a couple but i mean i'm talking about two out of like you know and this is already pre-filtered so i had already gone through and pre-filtered so i wasn't even looking at the like absolute garbage so this is like 60 or whatever resumes of like --- ### [AI Team Plans Need Token And Control Polishing](https://share.snipd.com/snip/445e0a4a-7726-4bae-8763-9acc074bcc1c) 🎧 43:29 - 48:16 (04:46) <iframe src="https://share.snipd.com/embed/obsidian-player/snip/d5e472d2-3d86-4adf-a575-8f7c69370e1f" width="100%" height="100" style="border: none; border-radius: 12px;" sandbox="allow-scripts allow-same-origin allow-forms allow-popups allow-clipboard-write" ></iframe> - Team AI subscriptions currently lack ergonomic controls like token pooling and clear rate vs weekly limits, creating friction for teams. - Ian moved the team to Claude/Codex and found rate-limit increases misleading and token logging/options immature. #### 💬 Quote > If I'm paying for three seats as a team, why are we not pooling tokens? > — Ian Landsman Ian Landsman on shortcomings of team AI subscriptions #### 📚 Transcript **Ian Landsman:** So this last week, I also moved everybody on the team over to Claude and Codex team plans. How's **Aaron Francis:** that going? That's freaking **Ian Landsman:** terrible. It's a nightmare. It's the worst experience in world. It's so expensive. I'm not even into, like, I've only crossed over into actually paying for extra tokens a couple times and i'm at like 10 extra dollars so i'm even saving money but it's like the freedom of like i don't have to think about the tokens is just gone on this and it's because they won't let you buy the 200 a month plan flawed let you buy 100 but even that seems kind of like nerfed a little bit. Although they did come out at end of the week when they did this whole XAI. They acquired a bunch of capacity from XAI. And so they doubled the usage limits and stuff. So that was probably better. **Aaron Francis:** Kind of. We'll talk about that in a second. Yeah, we can talk about it **Ian Landsman:** in second. But I don't know. It's just like you need the team accounts to do this properly because like there's just you need the team controls like everybody just on their own like this doesn't work like we had to let a guy go there a month ago i had to just like ask him to like turn off his thing like there's no i need to be able to like turn things off but like whatever and i don't know there's also this whole other weird thing so there's that side of it like the very businessy side of it. Like people, you hire people, you fire people. We need to be able to manage that. There's like, but then there's also just things like, I don't know. Like they're letting, like part of it with Impetus was like they let, Claude built a security tool that I would like to use. I thought it was for teams and enterprise it's only for enterprise so i still can't use it even though i'm on the teams so hopefully that'll roll out to teams soon um but they are they do seem to be releasing stuff to the teams at enterprise before the public on certain features not on everything so it's like that's kind of interesting. Yeah, I don't know. The other thing that's sort of weird is like, and I do feel like I don't like this, but it's like it has all the logging stuff, like enterprise logging stuff. Like I don't even want my own stuff logged. Like I just want to turn all that shit off. But you can't turn it off unless you're on the enterprise. And it's like, don't want to be on the enterprise because I just pay per token, which is way too expensive. So it's just a lot. I feel like they're doing all this stuff, shipping all these features. Yep. I need them to stop. Yeah. The same discussion I've given you. No, no, no. It's different than me, but yeah, I agree. them stop and I need them to like double back and like clean up some of what they've left behind them. agree. And just lock things in a little bit. Because I feel like it's a little scatterbrained right now on both Claude and Codex's team plans. And we need a little refinement phase there. **Aaron Francis:** Slow down for thee, but not for me. I agree with you for them. I disagree with you for me, obviously. Yeah, I agree. I think the, I have to imagine the logging is useful i know some companies have come out and said like hey listen we're paying for all your tokens those cannot be used for side projects and so i imagine that's that's helpful there i don't know i haven't i haven't seen what the logging even looks like so I don't know if it's helpful there. So to circle back on the five-hour thing, such a scam. It's not, it is, they didn't increase, like, how many tokens you can use per week. They just said the five-hour rate limits are increased, so you can just burn your weekly plan faster. And it got all this, like, it got all this applause for, like, oh, they increased their usage limits. They didn't. They just increased the rate limits. **Ian Landsman:** Yeah, but see, that's really important. That's actually the most important part to me. Because there's a lot of times where I don't use the whole week very often of limit. But I do a chunk like when i moved over to teams i was instantly like i did like nothing and i was like 20 of the five hour window and the problem is once you cross the five hour window you have a choice right like you can set it up to just stop working which i think isn't ideal or you can set it up to go into token using paid tokens and so that's where i'm going to use a lot of my paid tokens. It's like I have this block. Today's the day I can actually just code. I'm going to blow through the five-hour window, and now I'm going to be paying for tokens, even though I'm going to, in the end of the week, not have even used my full lot of tokens. All right, that is frustrating. So that part is super frustrating. I even think there should be like, I think all the tokens should be pooled. That's another thing. If I'm paying for three seats as a team, why are we not pooling tokens? Because not everybody's using it the same amount. Right? **Aaron Francis:** Great point. Yeah. --- ### [Security Work Delays Token Town Launch](https://share.snipd.com/snip/169fbbb7-37f0-44fa-87f5-7d37239c5b46) 🎧 53:57 - 56:22 (02:24) <iframe src="https://share.snipd.com/embed/obsidian-player/snip/1aee95b0-6e29-4afb-a35c-8489876ef342" width="100%" height="100" style="border: none; border-radius: 12px;" sandbox="allow-scripts allow-same-origin allow-forms allow-popups allow-clipboard-write" ></iframe> - Ian paused Token Town launch to focus on HelpSpot security and SOC2 work because AI makes automated exploit discovery easier. - He prioritized security hardening and logging to prevent automated AI-driven multi-step exploits. #### 💬 Quote > The AI is going to get really good at like not just finding an issue, but being like, oh, I can string 12 of these things together into an actual exploit. > — Ian Landsman Ian Landsman explaining security focus and delays #### 📚 Transcript **Aaron Francis:** The world is evolving. I didn't think that was what it was **Ian Landsman:** going to be. We to stop what we're doing and we have to do some heavy security stuff on help spot on surrounding supporting things for help spot because it's just getting too easy for anybody to be like hey ai like go in here and like hack me this thing and we at least have to have some semblance of you know we've done various sorts of security stuff over the years, but not in this AI era, the full deep dive. So as I talked on the show, built like AI security tooling, trying to use the Claude security thing if they ever let me in. I have used the Codex as a security thing, which I am in and have used, although the way it works isn't my favorite because it kind of like goes off PRs, which are commits, which is fine, but can only go back so far. So I have like an app that's older than going back so far. So it just won't look. It doesn't go back 20 years. Yeah, it won't go back. So like helping old code get checked, at least not directly. So blah, blah. So that's just it. So we've spent a lot of time on that. We spent a lot of time on getting, we kicked off our SOC 2 efforts to make Userscape SOC 2 compliant. Sounds miserable, but probably important. That's very But also important, also along the lines of the security stuff. So it's kind of like security vibe. Couple weeks, couple ones here. But that stuff's rolling. Plugged a lot of bugs. Fixed a lot of things. A couple real ones, a couple like on its own is probably not the hugest deal, but like this is one of the biggest fears about the AI era and security, right? It's like the AI is going to get really good at like not just finding an issue, but being like, oh, I can string 12 of these things together into an actual exploit right and so like so you kind of really are forced to fix a lot of these things that maybe in the past you would have pushed or maybe nobody would have found because it's like a weird internal escaping thing that we'd have to have like an ex privilege escalation first which requires all those other steps to even get to that point. But it's like, well, now like that's a thing. Like let's assume they're going to make it seven steps in. Well, I know you should always have assumed that the whole time, but whatever. We were living in the real world, but that was a hard thing to do. And unless you were selling a product that China was going to come try to hack you, probably you could survive without it. --- ### [Daughter Builds Horse Name Generator With AI](https://share.snipd.com/snip/8472774b-ecd1-432c-9d2d-cbb665e0ab18) 🎧 59:35 - 01:05:47 (06:12) <iframe src="https://share.snipd.com/embed/obsidian-player/snip/428e22d7-0240-4558-857d-a6d39c04dd4b" width="100%" height="100" style="border: none; border-radius: 12px;" sandbox="allow-scripts allow-same-origin allow-forms allow-popups allow-clipboard-write" ></iframe> - Ian and his 13-year-old daughter built a horse-name generator site using Claude, then published it on GitHub at MiaLandsman.com. - The project engaged his daughter, taught publishing basics, and produced a shareable app with favorites stored in localStorage. #### 💬 Quote > We have a horse named Jenna she was so excited she's like this is the greatest. > — Ian Landsman Ian Landsman describing building a horse name generator with his daughter #### 📚 Transcript **Ian Landsman:** ai story for you i **Aaron Francis:** love a feel good ai story last week **Ian Landsman:** me and my daughter we're having a little daddy daughter time she came downstairs talking about something whatever blah blah so i talk about horses ball walks into the horses as we all know update on the horse we still don't have a horse but of course you can follow up with that later **Aaron Francis:** but at least you're paying a ton of money no **Ian Landsman:** i'm not paying any money right now i'm not paying any money that's good i'm i'm only paying for lessons and stuff which is like normal money so yes but we will be paying a lot of money soon but we aren't yet anyway i was like hey let's like build an app to help us with this horse naming situation so we had a claw we use clawed and we had to build up like do the artifact so it could like show a html page and we're in clawed and we're going through it and then she was loving it she takes over the laptop she's typing it improving it a lot like now we got this big horse name generator that knows all the types of horses and all the stuff and then we went we went all the way with it we were like all right we're gonna put this on github and explain a little bit of that and then we're gonna like we already had her name registered so if you to mia landsman.com can see this website so we got the name pointed at the github the hosting and the thing and whatever and uh yeah so now there it is we have a horse named jenna she was so excited she's like this is the greatest um i love this she's sharing it with her friends all the stuff it was very cool uh and it was great because none of my kids like coding so this was a nice little. So **Aaron Francis:** this is your 13-year Mia? **Ian Landsman:** Yeah. This is the youngest. This is adorable. **Aaron Francis:** Yeah. Isn't it cute? Crimson Lady, Soaring Atlas, Twilight Willow, Jolly Apple. These are great. These are great horse names. Says how, you know, their **Ian Landsman:** breed and like how much they, how high they jump and all the stuff. And, you know, some different, depending on certain settings, like other things happen. So there's a little bit of that stuff going on. It was really cool. So I recommend everybody do that with the old kiddos when they get big enough. There's a to track your favorites. **Aaron Francis:** You can click the heart button and it tracks the favorites. Puts **Ian Landsman:** it in the local storage. So it's a cool little ai name generator thing i i feel like these guys should probably make a way that's a little easier to publish what you build um i agree that's not really in there right now so you still have to kind of know all this other stuff to do it but uh yeah it was really cool little horse name generator if you have a kid into horse names horses check out MiaLandsman.com. **Aaron Francis:** Well, this is really, this is feel good. This is really adorable. Yeah, it's so cute. This is not dissimilar, like obviously the tech is, but this is not dissimilar from how I got into coding. It was like, I want to do a thing. Let's build a little HTML page and then we'll throw it up. And then, you know, you go to school and you're like, I have a website. And the kids are like, you have a website? What do you mean you have a website? So presumably she's at, you know, she's at lunch and she's going to MiaLandsman.com and she's shown her friends that she built a website. This is great. I love this. Yeah, it's **Ian Landsman:** so good, right? This so pure. It's so pure. And yeah, that's the other thing. It kind of also is making me feel like, you know, people are worried about how you're going to have software developers and stuff. But it's like how people are going to get into it. But I feel like this is just like we used to have to do view source on a website we like and figure it out. This is just like a nicer version of that. Instead of viewing source on somebody else's website, I could build my own website. And they could do stuff that I find enjoyable. But now, like, if I don't have me sitting there, right, she would have to figure out, well, how do I publish this on a domain or whatever? And then that kicks you off on, like, how does that stuff work, right? And you just go down that same road we've all gone down to figure out how stuff works. But, you know, your entry's a little different. You don't have to struggle so hard to actually figure out how it works because the tool can tell you how it works. So, yeah, I thought that was pretty interesting. But, yeah, she got a big kick out it. Doing stuff with your kid like that is fun. **Aaron Francis:** Yeah. How was it being dad and getting to show her some visibility into what you do all day? Yes. This is a **Ian Landsman:** big problem I have. It's like the kids don't understand what you do. No idea. You're just on the screen. Like you're looking at the screen. What does that mean? Money comes in. We go do stuff. But the money's all virtual, right? Like, you know, they don't see money. They don't see the effort other than dad looks You don't come home **Aaron Francis:** covered in coal dust. Like, they have no idea what you're doing. And, like, they **Ian Landsman:** look at a screen all day too, right? Like, they're on the screens for school, for when they're home. And **Aaron Francis:** you're like, get off your screen, but I gotta go on the screen. Right. And it's like, oh, so they **Ian Landsman:** look at the screen, I look at the screen, that must be the same thing. So it's very interesting to like just give them just a little bit of insight into like how this works, what I do, even though it's not obviously direct, but it's like the website, how we create a website, blah, blah, what you might do with that. So that part's really cool. Definitely with the oldest, well, all three kids. The oldest, we did some HTML by hand stuff way back in the day. The middle kid, we like a, I actually did like a, what was like learn the code for kid things over the summer? Me and him did that. We were trying to build a game, which I actually think is a terrible starting point. I don't know. Everybody wants to have people build games. If you're really into it, if they're self-motivated to do that, I that's fine. But I think just starting there is a high bar. So that was fine, but didn't go super great. Never got a moment like this. This was by far the best moment of all of them in a coding task. So it was really cool. The AI facilitating the human interaction. Not all bad. Not **Aaron Francis:** all bad. Yep. So it was cool. So go to MiaLandsman.com to check out Mia's horse name generator. Subhead where every horse finds the perfect name. It's great. This is so adorable. I love it. Yeah, it's great. Tell tell me that your weird internet friend thinks it's awesome i **Ian Landsman:** will do and uh yeah that's the part that's cool too it's like it's actually doing something right it's not even just like oh it's my website and there's a picture of me up there it's like yeah oh it's like it's an actual application that actually does something that these kids think about a lot and stuff like that. ---