The Exchanges

Every argument clarity score on this site is built from rows on this page. Each question and answer was assessed with names hidden, the host's own answers included, on four things from 1 to 5: directness (does it answer the question asked), coherence (do the ideas follow), precision (concrete details and clear references), compression (says a lot per word). The weighted mix (30/30/25/15) is the exchange score. A person's published score averages their exchange scores on raw tape only, at least 8 of them, shrunk toward the cohort mean. Full method →

Josh Mohrer no published score: no usable exchanges on raw tape, and a fair score needs 8+ · coarse estimate ≈4.0/5 from 12 produced feed exchanges record → ← everyone

Every exchange below was scored with names hidden, four dimensions each from 1 to 5. An exchange's score is 0.30·directness + 0.30·coherence + 0.25·precision + 0.15·compression. The published score averages the raw tape exchange scores and shrinks small samples toward the cohort mean, so five great answers can't beat twenty good ones. Produced feed rows count only toward coarse estimates, never toward a full score.

clear all ✕
12exchanges match
0on raw tape
0redirected or not addressed
Answered produced feed D 5 · C 5 · P 5 · Cm 4 4.85

Q most people are, I know you've changed your flow. I've done some research on how you change your pricing. You've said in the past that it was too complicated, or maybe it was like Claude as I was analyzing it said that you shifted from complicated to simple. I imagine that most people are doing annual subscriptions with you, right? But that's where the majority of the revenue is anyway.

A It's where the most of the revenue is in terms of, like, I just looked at this, it's actually split a third, a third, a third, weekly, monthly, and annual in terms of people, but in terms of dollars, it's obviously way more unannual. Um, But like, yeah, I mean, what's so weird, there are some things about the app store that are just like the way things are. And had, and since I had never really done any work there before, I, I had been working for, for app companies, but not app subscription companies. And the whole idea of an annual subscription still makes no sense to me. I would never buy an annual subscription or anything. Like, I don't know what I'm going to like in a year.

AI assessment note: “split a third, a third, a third, weekly, monthly, and annual in terms of people”

Answered produced feed D 5 · C 5 · P 5 · Cm 4 4.85

Q That brings me to my sponsor. Zapier. I mentioned that they're a sponsor. You said I'm, I'm integrating them somehow into customer service. What are you doing with Zapier then?

A I'm integrating them into, so I get the, you know, you recorded wave and then you either read it in the app Or it'll send an email, or you can share it to Google Docs or Notion. And I built out a bunch of the share features myself on the tools, you know, OneNote, DocX, like Acrobat, like all the formats that someone would want. But I get inbound all the time, like, hey, I really would love for Wave recordings to land in my company's like CRM. It's this like esoteric thing that I never heard of. Do you have Zapier? And so I actually have a beta zap or zapier thing where once you set it up, every recording you do will trigger a zap and it'll land that wherever you want. And it's part of a bunch of things I'm releasing in the next few weeks that include API access. MCP access. So if you want to ask one of the other AIs about your sessions, you can do that now. So it'll be API, MCP, and Zapier are going to kind of roll on the same day as part of a like, take your data wherever you want to go. Um, I think there is value in being the, the app that facilitates making the data. So you're doing the recording, it's creating data. Everyone seems to like it for that. But it's like, then what happens? Like the next stage is what, you know, something has to be actioned. We're meeting. There's going to be some sort of output from that, either a to do list or like action items or whatever, or …

AI assessment note: “every recording you do will trigger a zap and it'll land that wherever you want”

Answered produced feed D 5 · C 5 · P 4 · Cm 4 4.60

Q That's, that's the idea you ended up pursuing. What are some of the other ones that you didn't? I'm trying to understand how you figured out what to do or what to play with.

A Yeah, I mean, that was the one that ended up being the thing. I had another demo that I built that I'm actually have, like, revisited, and I can tell you about why, but that would basically take New York State legislation from their API, and you know, the sort of idea, like, no one reads the bill. Like, the bill comes out, and it's too many pages, and it's too confusing, and there's too many bills, so nobody reads the bill. You know, I got I got involved in government at Uber. Um, you know, I don't, I wouldn't say that I was interested in it, but it, it was interested in me, and so I had to get sort of into the weeds on that stuff, and AI feels like a good way to make sure that everybody reads the bill, um, and that things can't be, you know, released in the middle of the night and then voted on the next day. Like, so I made this thing that would take every bill coming off the New York State Senate, uh, feed and summarize it, and do some other things like, like ask it, who's gonna care about this? Who is likely to support it? Who's likely to oppose it? Things like this. And so there's an idea there that I worked on with a buddy from Uber that we actually have revisited, and like I just said, I'll circle back on that and, and kind of explain why, but, um, we're gonna be releasing that as like a side quest in the next few weeks. Um, so that was one idea. I, I feel like Customer s…

AI assessment note: “I had another demo that I built that I'm actually have, like, revisited”

Answered produced feed D 5 · C 4 · P 5 · Cm 4 4.55

Q It is so beautiful to watch it get built in public. Let's close it out with the last segment of this conversation, which is how are you getting customers? And I wonder if you could take me through like the first customers, which sounds like it was your friends who were willing to give you feedback, but then the next and the next, and then what works for you?

A So right now it's, it's Apple search ads and Google ads for, for iOS and Android respectively. And I haven't been able to get meta ads to work in more than a year. Every few months I dip back in, I spend a bunch and it fails. And there's a bunch of reasons for that. Competition. My ads are stale. And I think most of those, those are both symptoms of a larger issue, which is that I'm just not that interested in doing it. Um, and I, every so often a few times a year, I'll bring on a new agency. We'll get up and running. They're like, we'll for sure be able to make this work because my metrics are pretty good. Like if someone installs, there's like a one in 10 that they're going to give me money. And that's like, I'm told is pretty high for this sort of thing. And so, like, so I need more installs, you know, and so I'll do between 800 and, and, and a thousand installs a day. And I think a lot of that is coming from, from app store search. You know, on, on the app store, I have a 4.9 star, like 11,000 reviews. So it just comes across as trustworthy. And on Android, it's getting up there, but not quite as good. So I, there's gotta be a word of mouth element that I'm not really measuring. There's gotta be some natural search, the ads and the, in those two places that are probably helping, but it's mostly, uh, yeah. And then really good retention. I mean, Well, actually not really goo…

AI assessment note: “right now it's, it's Apple search ads and Google ads for, for iOS and Android”

Answered produced feed D 4 · C 5 · P 4 · Cm 4 4.30

Q would only come on television and movies, and then eventually you end up with all these creators, and some of them are Mr. Beast, and some of them are doing it just for themselves because they like having family videos, and they're more creators. Ok, um, do you think this could be a billion dollar business? Do you think this is the Sam Altman one person billion dollar business direction?

A Well, I can also answer this the way that I always do, which is, so I think that it's usually framed as one person unicorn and unicorn refers to a company worth, you know, the billion dollar mark. Not, not no one will ever value this because I'm not going to raise money. So there's no way for this to be a unicorn in that sense, um, which is a lot of vanity and kind of like what investors are willing to pay. It's a great signal for sure, but it's part of a scheme that I'm not part of right now. Um, you know, I think what you're asking is like, what is the natural limit of something run by one person? And I think it's probably quite a bit more than what I'm doing.

AI assessment note: “no one will ever value this because I'm not going to raise money”

Answered produced feed D 5 · C 4 · P 4 · Cm 3 4.15

Q Let me pause for a moment on there because I don't think I've seen the ideas or understood the framework that you use to pick the ideas that you would pursue. I just heard that you had a bunch of little ideas and wave AI was one of them. What are they and how did you figure out what to, what the ideas would be?

A Sure, so I mean, none of it was sort of like, I'm gonna start a business. It was all just kind of like, I'm unemployed and have free time, and like, this, this is really fun. Um, but kind of right away, ChatGPT's, you know, the sort of like most amazing thing that it could do, in my view, was summarize things. So if you think about it, up until then, computers, and I've always loved them, The one thing they couldn't really do is summarize text, take a bunch of texts and make it a small amount of text. Like that struck me as the killer app. And in a lot of ways that still kind of is like, it really is amazing. We're totally used to it now. It's been three and a half years. It's part of our like day to day, but that is still pretty amazing that an LLM can read a bunch of text and make it a small enough. So one of the ideas was sort of like a two hop, like record audio, Transcribe that into text, which computers were already able to do at that point, and then take that text and summarize it. And the idea was sort of like, you're at lunch with a friend and you talk about all these great things and like, but then you forget.

AI assessment note: “one of the ideas was sort of like a two hop, like record audio, Transcribe”

Answered produced feed D 4 · C 4 · P 4 · Cm 4 4.00

Q run a marathon as a way of getting me to even run to 10 K. I'm wondering, do you have any number in your head that says, This is what I'm aiming for, even if it means that I'm not gonna run my version of a marathon and have a unicorn. I'm gonna figure out what the 10 K is that comes from that. What do you have in mind?

A Yeah, I mean, I think running, particularly as someone who got into it late in life, I had no, like, high school running record. So, like, I'm, I, in the last five years, have run as well as I've ever run, you know, half marathon, marathon, as well as I've ever done. Not good in the sense of all runners, but good for Josh Moore. So I, you know, it's related in that I sort of relish doing better than I've ever done in the past. And so I would like the, I would like the numbers for this year, both on my running and on the business to be best ever. Uh, last year was best ever business, not best ever running. Those things are related, unfortunately. Um, and when it's 20 degrees outside, I can't do much running anyway. But I, uh, I would just like it to grow. I think I have, yeah, there's no grand vision. I mean, it's, it did seven last year. I'd love it to do 10 this year. That would be great. I don't think I'll be able to do that, but let's see what happens. I'm just kind of enjoying the process. I'm unlike venture-backed startups, which have been exclusively where I've worked for the last couple decades.

AI assessment note: “it did seven last year. I'd love it to do 10 this year.”

Answered produced feed D 4 · C 4 · P 4 · Cm 3 3.85

Q level that I think they should. And that is recording in-person conversations. I want to be there and I want to summarize it. You got into that. How did you know that you were on the right track? Was it number of users? Was it, I know you got traction early. What was it that told you this is worth spending the next few years of my life obsessing on?

A I don't remember the actual answer, but the answer I give when I ask that is a lesson I learned at Uber, which is if someone on the support line is like exhibiting extreme emotion, Whether it's positive emotion or negative emotion, that is actually a good indicator that you're onto something. Like, if you can make a customer so mad that you think they want to murder you, you might be onto something there. I remember at Uber, we got a lot of that in the early days, both ecstatic happiness and, and, and, you know, serious rage. You know, the driver took a left on Fifth Avenue, and it, it wasted 20 minutes, and I'm gonna kill you. It's like, alright. We are, if you're that passionate, positive or negative about something, then that product has really hit. That's really there. There's value there. And so I think part of it was talking to users. You know, I still get a like, this is amazing. You know, for many users, if you're out of the bubble, you're touching AI, maybe not that often. ChatGPT once in a while, but you don't really know how to use it for your work. You're not in the bubble. And so this, for many of my users, I think, is their AI. It's what they use AI for. Do you use AI in your life? Yes, I use Wave. Like, I like to ask folks in my secular life, in my, like, not tech life, if they use AI yet. And you'd be surprised how many people say no. You know, yeah, for fun on …

AI assessment note: “if someone on the support line is like exhibiting extreme emotion”

Answered produced feed D 4 · C 4 · P 4 · Cm 3 3.85

Q be able to create, what does it mean? Does it mean that people are going to start creating their own version of wave? Does it mean that companies are going to start to do it? Or is, is it just that we'll always have both? We'll have what we have now at SAS. And in the same time, people will create their own software in addition. What do you think?

A I think there will be more software. So, I mean, wave doesn't have to be a team of 50 people venture capital in an office. It's me in the downstairs apartment in my building. I think there'll be more There'll be more software. I mean, I had this thought in the shower. I recently, I'm in New York, where we recently had a ton of snow, and I have young kids who I knew were going to want to sled, and I didn't currently have snow boots, uh, or pants for the snow. I left it upstate at my other house, and, uh, I bought on Amazon, like, very cheap Boots. You know, and they were great. They were like 40 bucks and they were great boots. Um, the pants were similarly great. So I felt like a hundred bucks on very, very useful things. And I wonder if software, it's like making apparel is like a solved problem. Like we know how to do that. We do it overseas. The supply chain is there. They make the boots really cheap in every size. They ship them. They go to Amazon. I, Pick the boots based on where I could get it the next day in my size, and I got them for really cheap. I sort of wonder if software might be similar in that it's cheap to make. There's a ton of it. None of it is, is huge. The boot brand is unknown. My kids and wife made fun of the way the boots looked. But boy, did they do the job. And so I wonder if software will have a similar kind of thing where there's just a lot of it. The…

AI assessment note: “I sort of wonder if software might be similar in that it's cheap to make.”

Answered produced feed D 4 · C 4 · P 4 · Cm 3 3.85

Q How, like, how do you shift from one to the other? What's your process and how are you using cloud code now? What's your setup in it?

A Well, the same underlying model, I'm still using opus, but I think Odd. You know, one of the things is you can buy the API by the month with that. So I spend 200 bucks and I think I get something like 3000 dollars worth of API usage from it versus going direct and using the API. I was spending 50 to a hundred dollars a day, which I'm happy to do because I'm getting serious value out of it, but I'd rather spend less money when I can. But also the agentic loops that it's put in The like, the way it does the model, you know, the model is the same, but the way it interprets and asks questions and does follow up and touches your files and your file system and bash and all this stuff is like really crazy. Uh, you know, just every day I'm having these experiences that blow my mind, and I'm running out of folks to tell about it, but it's like, I was talking about, I, I, like, initiated a discussion about my rates. Like, how much do I charge? Should I charge less? Like, Uber was very big about, like, time to cut prices. Like, we were, every six months, we were lowering the price, because that's a moat, you know, makes people, you don't want to get undercut, and I have no costs, and my, The margins are really good, and so I'd rather give some of that back. So I started a, like, a discussion, like, let's talk about this. Maybe I divide the features up and put one of them under a lower tie…

AI assessment note: “the agentic loops that it's put in The like, the way it does the model”

Answered produced feed D 4 · C 3 · P 4 · Cm 3 3.55

Q What about virality? Other note-taking apps. You do that.

A Well, I just, so in the early days, the only answer I could come up with, and there was a time where we did, it was a couple of years ago, where meta ads worked super well, and I grew from like a million ARR to like six in like four months, like it was crazy, because it was a first of its kind thing, and it just crushed it on meta. Now, there's forty million of them, so the ads don't really work anymore, but what I used to think, and I think this is still somewhat true, Is that if you're good, things might work out. I know it's sort of trite and maybe even not true, but I do kind of think that the internet's, the, the internet is good at servicing quality. Like it's good at servicing quality. You know, the app works. You know, and people seem to like it. They're bad. I get bad reviews. I get like, this app is buggy. And like, I might say like, you're using an iPhone, or using like, a Xiaomi Android phone or something, and it's like, oh, sorry. But.

AI assessment note: “meta ads worked super well, and I grew from like a million ARR to like six”

Answered produced feed D 3 · C 4 · P 4 · Cm 3 3.55

Q What's the Microsoft tool that lets you see how people are using your app? You're not using something like that.

A Yeah, I am. I have, I have century that like reports every problem that anyone has, whether they know it or not. But like, you know, and then I have a sort of a circle of people that I know from the real world who I gave free access to for forever, and one of them sent me a text last night. I was in bed. She used to work for me at Uber, and what time was this? I'm just looking at the text. It was 9:15 p.m. I guess I hadn't seen the text until I was in bed. I get a lot of sleep, by the way. I was in bed, and I saw a text, and she's like, did you change something on the, like, on the lines facing of the summary? Because it's making me go, I was like, oh, you're right, I did. Sorry. And then I fixed it. Um, but there's just, like, little, that would never come up in, like, a, a bug report. I, I changed how Markdown is rendered in the app for a reason that's not that interesting. But yeah, the, Line spacing. And I even noticed, and it's just like, when you work alone, I like notice things and don't, oh, and, but it doesn't always register, because there's no other voice, you know. So it's very, very useful to just have people in the orbit saying things, and in, in the absence of people I know, I rely on support channels.

AI assessment note: “Yeah, I am. I have, I have century that like reports every problem”

page 1
Made with StarZero

Turn any episode into a week of clips.

This entire site, about 40 episodes transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.