On this episode of AI & I, Dan Shipper interviews Sam Gerstenzang and Dan Friedman, partners at Boulton and Watt, a self-described "world's slowest startup incubator." Their model: come up with an idea, grow it to $5–10 million in revenue themselves, then hand it off to a CEO to scale. Companies built with this playbook include Moxie, a Series C company helping nurses open their own medical spas (600+ customers, 200-person team globally), and Meadow Memorials, a funeral home company with no physical real estate that has become the largest provider of funeral services in California. Both launched around the arrival of ChatGPT without being built for AI. The conversation covers building an AI agent called Matthew Bolton to power customer discovery, why synthetic customer calls completely failed for them, deploying AI inside established companies, why newer projects see bigger AI gains than mature ones, and why they believe you shouldn't give anyone credit for using AI. # Channel Presentation Observations: "AI and I" (Dan Shipper) ## Observable Channel Identity - **Hosted by Dan Shipper**, who introduces guests and reads sponsor copy in first person ("Dan here and I want to take a second away from the episode..."). - The closing outro names the show: > "smash that like button and subscribe to AI and I" — an explicit channel self-identification, delivered in a comedic, self-aware tone (the outro ends with "I'm absolutely hopelessly in love with you"). - The host is identified as running **Every**: "I'm in the Every office uh in Boerum Hill" and "We like to move really fast at Every. We're always shipping new features and we run five products internally with just 20 people." ## Claimed Scope (from creator statements) - Focus on **AI, ChatGPT, and AI-driven work practices**, per the outro: > "pure unadulterated knowledge bombs about Chat GPT." - Interview format featuring **founders and builders** — here, two incubator founders (Sam and Dan of Bolton and Watt) discussing AI strategy, AI-augmented company discovery workflows, and AI adoption inside portfolio companies. - The host engages as a **peer practitioner**, not just interviewer — he references his own product work: "cuz you're building new products all the time" and "my experience we do a lot of like AI transformation type stuff with big companies." ## Intended Audience - **Startup founders, operators, and AI-curious knowledge workers** — evidenced by episode content: AI tooling for customer discovery, agent orchestration, AI adoption inside companies. - **People evaluating AI tools for real work**: the host's framing questions like "Have you guys seen any change since Opus 4.5 came out?" and "how are you thinking about how the model evolves now that AI is a thing?" ## Monetization / Sponsorship Presentation Three mid-roll sponsor reads read personally by the host, each tied to his own workflow: - **Grainola** (AI meeting note-taker): "I use it pretty much every day... It acts a little bit like a leadership log for me" (promo code "every"). - **Intent by Augment** (agent-orchestration developer workspace). - **Framer** (website builder): "Non-coders on our team use it all the time to ship website updates" (framer.com/dan). Pattern: sponsors are **AI/productivity/software tools**, and reads are framed as host-verified usage rather than generic placements. ## Demonstrated Presentation Patterns (not claimed) - **Long-form conversational interviews** with a Brooklyn/NYC tech-scene peer network (guests described as "good friends," prior shared history in the 2012–2013 NY tech scene, a meditation retreat, offices "a few blocks away"). - **Recurring guest invitation model**: "whenever your new thing launches, I would love to have you back." - Host's own stated product philosophy surfaces in dialogue: > "I think of new products as content first and then in businesses second" and "my products start as basically blog posts." - Tone is **casual and humorous** (laughter markers throughout; the absurdist outro), mixed with substantive operational detail. ## Distinguishing Claims from Demonstration | Type | Evidence | |---|---| | **Explicit self-description** | Outro naming "AI and I," Dan Shipper, ChatGPT knowledge focus | | **Demonstrated pattern** | Founder interviews; host-verified sponsor reads; peer-level AI practitioner discussion; Brooklyn tech network framing | # Delivery Observation: "Meet the Slowest Startup Incubator in the World" ## Format Overview - **Three-party conversation**: host Dan Shipper interviewing two guests (Sam and Dan, of Bolton and Watt), with speaker attribution marked by `>>` and intermittent `[laughter]`. - **Podcast structure with sponsor interruptions**: the conversation is broken by three ad reads (Grainola, Intent/Augment, Framer), each re-entering with a variant of the same phrase: - > "And now back to the episode." - > "And now, back to the episode." - **Closing oddity**: the transcript ends with a highly exaggerated outro ("smash that like button... pure unadulterated knowledge bombs about Chat GPT") whose register clashes sharply with the rest of the discussion—possibly AI-generated or parody, but this cannot be confirmed from the transcript alone. ## Tone ### Conversational and self-deprecating - Guests repeatedly downplay their own expertise: - > "Sam and I this morning looked at each other and we were like, are we sure we should be on this podcast?" - > "we're not like 30 days from launch and announcement. Um we're we're a few months out." - The host mirrors this casual register: > "This is kind of sick. I love this." ### Animated, humor-laced - Frequent `[laughter]` markers (at least six instances), often following jokes: - > "All the smart people are listening to Dan and Avery, and we'd rather compete against the less smart people" - > "All right. RIP John Henry." ## Vocabulary & Recurring Phrases ### Investor/startup jargon is pervasive - **Repeated framing devices**: "AI native," "AI durable," "business in a box," "zero to one," "greenfield," "point of view" (abbreviated "POV"), "hypothesis tracker." - The AI taxonomy phrase appears nearly verbatim twice, suggesting it is a rehearsed talking point: - > "There's the AI native company that pushes the ball forward inside of some category or there's the AI durable company that effectively uses AI where the core of the machine is not going to change." (appears at [12] and again at [906]) ### Filler and hesitation patterns - All speakers rely heavily on **"like"**, **"you know"**, **"um/uh"**, and **"sort of"**: - > "we sort of her back office that helps her stay compliant" - > "I think it's like a useful distinction to be like okay what are like things that are like purely enabled" - Repeated restarts and self-corrections: > "you shouldn't give anyone credit for using AI, but you should make sure that the expectation is that they use AI, or sorry, the expectation is they'll deliver the best product" ### Vivid metaphor as a teaching habit - Both guests explain concepts through physical or narrative imagery: - > "eat the glass figure it out you know push the boulder up the hill figure out if the boulder wants to go uphill" - The extended **John Henry** digression as an analogy for human-vs-AI research. - > "Potemkin villages of like I used AI for all this stuff and it's just like it's all just for show" ## Teaching / Explanatory Habits ### Concrete example after abstract claim - Guests consistently anchor ideas in specifics: - After describing the incubator model: > "Our first company is called Moxie and it helps nurses open their own med spas." - On internal AI tools: > "a designer spun up an app where anyone can type in a hospital name, a town name, and it resolves whether we can find it or not." ### Numbered / two-part structures - A recurring habit of splitting ideas into two contrasting buckets: - "AI native" vs. "AI durable"; "greenfield" vs. "existing systems"; "research development and then there's like production." ### Walkthrough / show-and-tell - When asked about their AI workflow, one guest pivots to a live demo-style description: - > "we ask it to reread our latest point of view, we ask it to read the hypothesis tracker, read the most recent 10 calls..." ## Communication Patterns ### Host as active synthesizer and challenger - Dan Shipper restates guest points then probes: > "It sounds like you have like uh the greenfield things are quite fast and then anytime it has to touch something already exists..." - He pushes back with personal experience: > "my experience is if you don't... then people will just continue doing it the way they already know how." ### Guests finish each other's frames - The two guests alternate answering, explicitly building on each other: - > "And And somebody that's the role like Dan and I play for each other like is keep each other honest." - > "The only thing I'll add is..." ### Honesty about failure modes - A distinctive pattern: guests volunteer what *hasn't* worked, at length: - > "We desperately tried to have it be like in sincerity mode and we can't get it out of that." - > "we just can't make that. We haven't yet been able to make that work at all." ## Uncertainty Notes - Speaker attribution is unreliable—`>>` markers don't consistently distinguish Sam vs. Dan (the guest), so per-individual patterns cannot be separated with confidence. - The final outro's jarring tone shift is observable, but its cause (parody, generated text, or sincerity) is **not determinable** from the transcript. - Emotional states like "existential dread" are mentioned *by the speakers themselves* as quoted content, not inferred by this observer. # Observation Document: "Meet the Slowest Startup Incubator in the World—Pumping Out Billion-dollar Companies" ## Participants - **Host (Dan Shipper, "AI and I" podcast)** — identified via the outro reference "Dan Shipper as the captain of the spaceship" and his role at Every. - **Sam and Dan (guests, co-founders of Boulton & Watt incubator)** — the transcript labels one guest as Sam and one as Dan; speaker attribution between the two guests is inferred from context within the transcript (e.g., "Dan stayed on as CEO" of Moxie, Sam started the second business) and is noted where uncertain. --- ## 1. The Boulton & Watt Model **Guest (Sam, per context):** The incubator starts a new company "every couple of years," in niche verticals combining software, services, and a real-world component. They run each company to **$5–10M revenue**, then recruit a CEO "who's better than us" to scale it, remaining deeply involved as long-term board partners — contrasted with traditional incubators that hand over money and coach "from the sidelines." - Two companies started to date; a third is imminent. - They self-describe as "the slowest incubator in the world," though they're exploring accelerating to one company every 18 months, or a year. The two constraints: **quality of idea** and **talent** (early employees, executives, CEOs). **Rationale offered:** - **Guest (Sam):** The hardest, highest-value skill is the early product-market-fit phase — "eat the glass… figure out if the boulder wants to go uphill." He credits the model's success partly to luck: "if it didn't work people would be like oh your model doesn't make sense." - **Guest (Dan):** In his previous 10-year company, "years four through ten" involved "a constant existential question" about whether he was spending time well. The incubator structure lets him avoid — or distribute across five companies — responsibility for that phase. He qualifies: "I'm sure it will have some other… psychological challenges." - **Guest (Sam):** Founder skills differ by stage; specialization and repetition at zero-to-$1M revenue is itself valuable. **Work division (Guest, Sam per context):** The two "tag team" — one runs the current company while the other starts the next. Sitting next to each other, they maintain full mutual context, act as thought partners, and retain "all the upside from co-founders" with larger scope. A newer partner (Wes) specializes in concept development so founders "roll into" a validated idea rather than a cold start. --- ## 2. Portfolio Companies - **Moxie** (started ~3.5 years ago): helps nurses open med spas; back office for compliance, growth. "Hundreds and hundreds" of clinics; **Series C, tens of millions in revenue, 600+ customers, ~200 team members**. Guest notes it launched shortly before ChatGPT and has had to *adapt* to AI rather than being AI-native. - **Contemporary funeral home**: no physical real estate; arrangements online/phone; in-person events at wedding venues idle on Tuesday mornings. Claimed to be **"the largest provider of funeral services in California"**, launching in other states. **Guest (Sam) on why these businesses:** They're hard to build against (operational complexity requiring simultaneous services + software company building), with fewer competing startups — not "10 YC companies starting each of these." --- ## 3. AI Strategy: "AI Native" vs. "AI Durable" **Guest (Dan) framework:** Two good company types now: 1. **AI native** — pushes the ball forward in a category. 2. **AI durable** — uses AI where "the core of the machine is not going to change." Boulton & Watt positions in the AI-durable category. Examples of reasoning: - No "robotic injector" expected for 7–10 years; what happens inside a med spa isn't deeply affected by AI. - Every AI-native idea they've evaluated had "multiple formidable looking competitors doing something almost exactly like our idea." - **Host joke / guest response:** All the smart people are doing AI-native; guests would "rather compete against the less smart people." **Host's probing note:** He recalls Dan having said "nothing's going to stay the same, it's all totally going to be different" in earlier conversations, and observes Dan may feel differently now — Dan's current position (as expressed here) is the durable/AI-as-accelerant view. This tension is preserved, not resolved. **On naming:** Boulton & Watt is named after the 1775 steam-engine commercialization company — hyper-aware of technological change while deliberately counter-positioning on "what's going to continue to be valuable." **GEO/generative commerce:** Guest reports only incremental change — more traffic from ChatGPT treated as another paid-search-like channel; "we don't think people would buy a funeral via chat." Acknowledges this may not generalize to other products. --- ## 4. AI in the Company-Discovery Process **Guest (Dan, per context) on the dichotomy:** Seed-stage founder friends report "engineering is 10x faster"; Series D friends report "~10% faster" — "the classic dichotomy right now." **What has worked:** - Vertical discovery (formerly a week of Googling) is now mega-prompts to generate and assess categories. In one exercise, the group selected three categories — one from the human "point of view," two where the AI was "screaming" yes. - Forced migration from Google Docs to Notion. - **"Matthew Boulton" agent identity** (named after the actual Boulton of Boulton & Watt): preps calls by reading persona, hypotheses, and validation focus; post-call, ingests transcripts and re-updates evidence for/against each hypothesis with supporting quotes. - Reported as "a huge hit" for call prep and hypothesis tracking. **What has NOT worked:** - **Synthetic customer calls** (AI playing the customer): "We haven't yet been able to make that work at all." The AI expresses "a 10 out of 10 customer poll" for anything passing a basic sniff test; attempts at "sincerity mode" failed. They flew to meet a real prospective customer and "fell on our face. Completely." - Conclusion (Guest): AI "can help us talk to people effectively, but it cannot actually reduce the number of people we need to talk to to get to confidence." **Guest epistemic humility on Matthew Boulton:** He does *not* trust it with high-level opinion questions ("I take those results with skepticism") but finds it "really good at finding the quotes" supporting/refuting hypotheses. AI also serves as a devil's advocate — "make the best counter argument" to sharpen thinking. Sometimes the counter-argument stings: "Oh, off. Like, what are you saying?" **Guest (Dan) on research quality:** AI beats "low and medium quality reporting" and is "quite a good parser" of thick PDFs and public company filings. But: - **Host's claim:** He trusts AI for consensus opinion on a space, but trusts the guest to return something "new and interesting… that I don't think that Claude can get to on its own." - **Guest's response:** Agrees — "I don't think we've gotten good ideas from it." Ideas come from an "earned point of view" (e.g., ~60 conversations with "business in a box" founders) fitted with AI-consumed facts. Asking Claude for business-in-a-box ideas directly "wouldn't be ones that felt good to us." --- ## 5. Bolting AI onto Existing Companies **Guest (Sam, per context) on AI initiatives:** - A formal "AI initiative" is a **bad idea** — it "leads with the hammer," analogized to the NoSQL and Slack-bot waves. - But something must "shock the system." - Operating principle: **"You shouldn't give anyone credit for using AI, but… the expectation is they'll deliver the best product and output knowing that AI exists."** No credit for obviously AI-generated bad copy; full credit for prompts that get 70–100% of the way there. **Host's counter-experience:** In big-company transformations, just stating that expectation means people keep doing things the old way, since that's how they know to get best quality. **Guest's remedy:** Find a few people to set examples, publicly compare work, and "raise the floor" — rather than counting AI-generated PR percentages (which invites blaming AI for bad output). **Specific wins:** - Landing-page generation — though integrating back into Webflow with consistent headers/footers is substantial work; a two-phase research/production split. - **Throwaway apps**: e.g., a designer built an app resolving whether the funeral business services a given hospital/town, deployed as a separate link rather than integrated. - **"SamGPT"**: trained on Sam's blog posts, used by the recruiting team for LinkedIn outreach in his voice — "I forgot it existed, and that's worked really well." - Engineers are faster at coding, but "deployment still takes work, maintenance still takes work." **Host's synthesis:** Greenfield work is fast; anything touching existing systems is accelerated but not transformed. Guest: "that's exactly right." **On Opus 4.5:** Early read from Moxie engineering is "better but not a step function different experience," for a 40-person team on a mature codebase — not the "night and day transformation that the X sphere is reporting." --- ## 6. Next Business (Unrevealed) - **Guest (Dan):** Declined to share — not secrecy so much as "the level of embarrassment at literal day zero"; they're "a few months out" and only decided "as of the last five days." - **Guest (Sam) on criteria:** Looking for the intersection of AI with other secular trends (e.g., rising death rate, demand for Botox) rather than AI as the primary trend; P&C insurance is one area under exploration, with some transformation aspects and some that will work the same. --- ## 7. Closing Contrasts Between Host and Guests - **Host (Dan Shipper):** Ships products as content first — blog posts of weekend experiments — businesses second, "if they seem to have legs." - **Guests:** First question is "will someone pay for this?" — "we're both doing something kind of weird, but in really different directions." - Historical parallel noted by host: the guest previously ran an automotive professional newsletter business (pre-"built in a week" era), which "ended up not working" — guest acknowledges it "maybe executed a different way would have worked." --- ## Unresolved Tensions (Preserved) 1. **Dan's earlier vs. current AI stance** — host explicitly flags a past "everything changes" position vs. the present AI-durable stance; the shift is acknowledged but not explained. 2. **Luck in the model's validation** — Sam concedes the model worked "the first time" partly by luck, which undercuts the "we've done this two times" proof claim he simultaneously makes. 3. **Expectation-setting vs. inertia** — host's enterprise experience contradicts the guests' expectation-based approach for mature organizations; the guests' fix (examples, floor-raising) is asserted, not tested at enterprise scale. 4. **Guest uncertainty** about their own future: "maybe we'll end up changing our minds and finding the other [AI native] category is really fun." ## Note on Speaker Attribution The transcript uses "\>\>" turn markers without explicit names. Attribution of specific quotes to Sam vs. Dan relies on contextual anchors (e.g., "Dan stayed on as CEO" of Moxie; "SamGPT trained on my blog posts"). Where the anchor is absent, statements are attributed to "Guest" or the most contextually supported individual, and readers should treat fine-grained guest attribution as provisional. # Speaker Attribution & Provenance Record **Video ID:** `01a0f15a-a93a-7b74-a7b5-511e32b18d8e` **Title:** *Meet the Slowest Startup Incubator in the World—Pumping Out Billion-dollar Companies* ## Speakers & Roles - **Host — Dan Shipper** (identified via sponsor read "Dan here" and outro naming "Dan Shipper as the captain"). Runs **Every**, records from the Every office in **Boerum Hill, Brooklyn**. Serves as interviewer throughout. - **Guest — Sam** (co-founder, **Boulton & Watt** incubator; transcript spells it "Bolton and Watt"). Discussed as co-founder of Moxie and the funeral home business. - **Guest — Dan** (co-founder, Boulton & Watt; first CEO of **Moxie**; was previously at "Emigre and A16Z" per host). - **Unnamed sponsor-read voice(s)** — deliver mid-roll ad reads (identified only as "Dan here" for the first; the others are unattributed in the transcript). - **Outro promo voice** — delivers the closing like/subscribe pitch for the show **"AI and I"** at the end. > Note: The transcript uses "\>\>" markers without reliable speaker labels; attribution of individual lines between Sam and guest-Dan is based on explicit in-text cues (e.g., "Sam and I," "Dan stayed on as CEO, I went to start the second business") and remains partially uncertain at line level. ## Promotional Segments (Flagged) 1. **Grainola** — AI meeting note-taker; promo code `every` for 3 months free (`grainola.ai/every`). Mid-episode break. 2. **Intent (Augment / Augmentcode)** — developer workspace for orchestrating AI agents; `augmentcode.com/intent`. Mid-episode break. 3. **Framer** — website builder; `framer.com/dan` for 30% off Framer Pro annual plan. Mid-episode break; notes Perplexity, Miro, Mixpanel as customers (third-party claim within ad copy). 4. **Closing promo** — like/subscribe call for "AI and I." ## Entities & Ventures Referenced (per guests) - **Boulton & Watt** — incubator; named after the 1775 steam-engine commercialization company (guests' claim). - **Moxie** — Series C med-spa business for nurses; guests state tens of millions in revenue, 600+ customers, ~200 team. - **Contemporary funeral home** — guests claim largest funeral provider in California; no physical real estate. - **Matthew Boulton** — internal AI agent identity (Notion-based) used for customer discovery; used in a P&C (property & casualty) insurance category dive. - **Sam GPT** — internal tool trained on Sam's blog posts for LinkedIn candidate outreach (guests' claim). ## Third-Party References - **YC / Y Combinator**, **McKinsey**, **Notion**, **Google Docs**, **Webflow**, **Claude / Anthropic (Opus 4.5)**, **ChatGPT / OpenAI**, **Augie**, **Codex**, **Open Code**, **GLP-1s** — all mentioned by speakers descriptively. ## Availability & Uncertainty - **Guests' next company:** explicitly withheld — "we cannot" give a preview; a few months from launch (guests' statement). - **No curated metadata was supplied** with this video; all provenance above derives from the transcript. - **Comments availability:** not stated in source. I love this. \>\> Oh, thanks. I feel like that's actually really meaningful. You know, I got to say Sam and I this morning looked at each other and we were like, are we sure we should be on this podcast? \[laughter\] There's two good companies to start now. There's the AI native company that pushes the ball forward inside of some category or there's the AI durable company that effectively uses AI where the core of the machine is not going to change. \>\> \[music\] \>\> Dan here and I want to take a second away from the episode to tell you about Grainola. Grainola is an AI note taker for your meetings and I use it pretty much every day. That may sound a little bit weird or a little bit creepy like transcribe all your meetings. Well, for me it's actually kind of indispensable as a leader. Every is about 20 people now and it's really important to me that I understand how decisions get made, how I'm showing up in meetings and how I can help my team the best way I can. Grainola acts a little bit like a leadership log for me so I can see how I've done in meetings, what situations came up in a particular week and how I can do better next time. If you're trying to improve as a leader and scale your company, try Grainola as your AI powered notepad for meetings. Head to grainola.ai/every code every to get 3 months free. And now back to the episode. Sam and Dan, welcome to the show. \>\> Glad to be here. Thanks, Dan. \>\> Thanks for having us. \>\> Uh you guys are both uh good friends of mine. You run the incubator Bolton and Watt which I think is one of the most interesting incubators um that I've ever run into and you're coincidentally maybe or maybe not so coincidentally you run it down the street from us. So, I'm in the Every office uh in Boerum Hill and I believe you guys your office is like a few blocks away, right? \>\> Yeah, it is. Yeah. \>\> Um and uh uh Dan, we went on a Jhana meditation retreat together uh a few a few months ago. So, there's just a lot of like interesting overlaps um and I just really respect you guys. I think it's very easy to say, "Oh, I'm running a startup studio." and it's very hard to actually do it well and you're one of the few people that do it well. \>\> And it's funny I was I was thinking before this I met both of you like 13 years ago in the New York tech scene. It was like 2012 2013 and so we were sort of like in the same scene for a long time and then Dan Dan Freeman and his business partner Emma I reconnected with at a party you hosted maybe 10 years ago. \>\> I didn't realize it was at my party. \>\> It was at your party and so you were like you know interwoven part of of this story in these funny ways too. \>\> I take all credit. I don't know why I don't get Carrie like \>\> \[laughter\] \>\> That's amazing. That's that's really fun. Yeah, I remember you were at like Emigre and A16Z and yeah, very yeah. Okay, well I guess maybe let's start with for people who don't know what you guys do and what it means to be I think you call yourselves the slowest incubator in the world. What that actually means and what how your model works. \>\> Yeah, so we try to start a new company every couple of years often in like a really niche vertical that somehow combines software services and some real world component. And the idea is we come up with idea we run ourselves through five or 10 million revenue and then we go find a CEO who's better than us to take it from the sort of the next next 10x. And we remain involved as a partner to the company for its life really involved board members who have spent thousands of hours thinking about the whole competitive landscape, the company, the competitors, all all this stuff. So it's a really different relationship than a traditional incubator which may say, "Okay, here's a million bucks. Here's an idea. I'll help coach from the sidelines." but we're actually like in the seats and that means we're really concentrated. We've started two companies to date and we're about to start our third. Just to give you an example of the types of things we like to do. Our first company is called Moxie and it helps nurses open their own med spas. So these are nurses who are doing aesthetic medical aesthetics like Botox, filler, lasers, etc. And we're sort of her back office that helps her stay compliant, grow her business and really everything you need to do with med spas. We now have hundreds and hundreds of these med spa clinics across the US all in partnership with the nurses. So that's our first business we started about three and a half years ago. And then our second one is a contemporary funeral home. And what that means is we have no physical real estate whatsoever. We do arrange everything online over the phone. When we have sort of in person funerals, they're generally at wedding venues that are booked out a year in advance Saturday night but totally open Tuesday morning at 9:00. We're now the largest provider of funeral services in California and just about to launch in other new states. So we have a taste for these sort of like weird businesses that are like not YC zeitgeist that have real world applications and are the sort of reimagining these types of bundles. \>\> I I love it. Wait, tell us where is Moxie stage wise? \>\> Sure. So Moxie is a series C company you know into the into the tens of millions in revenue and 600 plus customers. Team globally is like 200. So it's like a comfortably mid stage company. And vis-a-vis AI we launched it you know some number of months before the release of chat GPT. And so it is hilariously a sort of like just before AI company that had none of that in in its conception and you know it has had to almost as much as a company that was started in like 2015 has had to sort of adapt as opposed to being much more native to that way of thinking. \>\> And so your model is like okay I'm going to do all the really at least for me the really fun stuff which is like I'm going to come up with the idea and then I'm going to do all the like hardest stuff of just like rolling the boulder up the hill for years until it starts rolling by itself and you're like I'm I'm out. Like how does how did you come up with that? Why do you do it that way? How is the whole thing structured to make this work? It's very it's a very like weird different thing. \>\> Yeah. I mean I think to some degree it's like an intersection of where we thought it'd be the most fun and also we thought like the most value would be created. I think like we looked at a bunch of other incubators and I'll put every aside for a second but I think there was sort of this like someone who wanted to help work ideas and like let those go and search see and we realized I think to make to maximize success of every shot we realized that we needed actually go you know eat the glass figure it out you know push the boulder up the hill figure out if the boulder wants to go uphill as Dan often will will say about that sort of early product market fit journey. Um and so we thought that was the place where okay if you can if you could ask someone else to do that you could like pay McKinsey to come up with startup ideas but the hard piece is like figure out like okay what's actually market signal how do you make changes when no data quite tells what to do and if we could get really good at that we get permission to all kinds of other things as well. Um and so that that was sort of like the origin of it and I think I think Dan and I both feel really lucky that it works so well the first time because I think if it didn't work people would be like oh your model doesn't make sense and now we can be like oh actually it has we've done this two times, and we're going to keep um doing it in that way. Um \>\> The only thing I'll add is and and maybe the years is a little bit of a gross heuristic, but my experience, and I think a lot of my founder friends' experience, is uh is pre- before to this, I built one company over 10 years, and years four through 10, there was a little bit of like a constant existential question of am I doing the thing that is like most interesting and most useful, and am I spending my time the right way? And I sort of like learned the physics of the business, and now I feel like I'm in purely execution mode, and um that feeling of like a little bit of existential dread, one way to avoid it is simply to not be doing not being directly responsible for years four through 10, or to uh vice versa, be responsible for years four through 10 across five companies, you know, by the time we get there. And uh I think that I'm sure it will have some other challenges that uh you know, some other psych- logical challenges, but um you know, it we are we are I think for me at least, I feel that we're we're here optimizing for like what's most enjoyable and exciting for us. \>\> We also sort of like pretend on the founder journey that's like the is the same skill set all the way, um and I think like what you do and how you do it really depends on the stage of the business. And so, there's actually like a lot of value specialization, too. And like, okay, we we know what this looks like going from like zero to one million revenue over and over again, and here's what takes from one to 10, or the systems, or the people, um how do you build the intuition? And so, we're also just like getting a lot of reps in in that early stage that very few people actually get to do that and see what success looks like on the other side. \>\> You guys also have an interesting model for how you break up the work between yourselves. How does that work? \>\> Um it's it's changed a little bit over time. Um the like when we So, when we started, we both started uh Moxy together, um and then Dan stayed on as CEO, I went to start the the second business. And at least for now, we're sort of like tag team it. So, I'll run one business, Dan Dan starts the next, and then I'll sort of start the the one after. Um, we'll see how long that lasts. I think one of the things I've really um appreciate about this working model is Dan sits next to me. We each have full context of the the thing the other person's working on. We can be really great thought partners and push the other person and emotionally supportive when things aren't working. Um, and also we have um our own space to try things our own way at the same time. And so, I've found that really um a useful way to build companies where you have all the upside from co-founders, um but also like are just a much larger scope and span, too. \>\> And then we have other partners inside Bolt and Watt, some of whom work with us within the company, and then we have one who has been with us for 4 or 5 months who specializes in the concept development phase. So, Wes generated his validated ideas. And the the thesis there is historically, when Sam and I rolled off one company, we rolled into a cold start on the search for the next one. And going forward, we want to be rolling into we've just validated our next idea and we're we're pressing go. Um, and so we're starting to like bit by bit, not in a like overly accelerated way, um institutionalize and specialize in the different components that we want to be best in the world at. \>\> And I I think we realized like the the two constraints that task are speed. Like, we we won't be the world's slowest, but we also could be a little faster. Right now, we're like \[laughter\] one every 2 years. And so, like the question we asked ourselves last year was like, could we take it to 18 months and what would that look like? And maybe take it to a year. Um, and the two constraints, like the first point that that Dan said, like, do we have a great idea? Um, and do we have a great one once a year? And I think we are extremely big believers in execution and also believe the idea matters a lot. And so do we have a good idea? And then second, our other biggest constraint is just talent. We need to find great early stage employees, great executives, great CEOs. And those are sort of the two things that are holding us up back and sort of limit our speed. \>\> And okay, so now it seems like and you tell me if I'm getting this wrong, but it seems like a lot of the the kinds of opportunities you guys gravitate toward are like real world like unsexy business like funeral homes and you know, med spas and at least with Moxie it's sort of a business in a box type thing where you you give someone the tools that they need to start one of these businesses and you partner with them and maybe take a cut of the revenue or something like that. I don't know exactly how it works, something like that. And I'm curious how that model or how you're thinking about that model in an in an AI world. You started Moxie right as ChatGPT came out for the first time. Yeah, how how does that how are you thinking about how the model evolves now that AI is a thing? \>\> Yeah, I think we were first attracted to those businesses because we thought they would be really hard to build against and there's a huge existing opportunity. And we said, "Okay, we're pretty good at this sort of operational complexity where you might combine you almost have to build like a services company and a software company at the same time. The sort of competition looks a lot different. There aren't you know, 10 YC companies starting each of these. We sort of like that space to play. Um and then I think one as sort of like ChatGPT came out and the world started to change, I think one of the things we realized in a kind of accidental way is the types of businesses we're spending time on are maybe more resilient to AI trends. Um, and we should think about AI as like an accelerant, of the the sort of speed in which you can build these businesses, but fundamentally like the business model of a funeral home, or a med spa doesn't change because AI's out there. Um, and the other day I was thinking back we so we named Bolton Watt after a company that was formed to commercialize a steam engine in 1775. And so we're like hyper aware of like how technology changes businesses. And also at the same time I think we've chosen maybe a sort of counter position to say, "Okay, a lot of things are going to change, what's going to continue to be valuable, and how do we sit with the current on those?" \>\> And how does that work in your mind Dan? Cuz I we've had a few more existential conversations about um, where AI may go. And I and I at least at certain times in our conversations you were like, "Nothing nothing's going to stay the same. It's all totally going to be different." And I don't actually I think you're maybe a little bit diff you're feeling a little bit differently now, but where where are you currently and how does that filter into the strategy? \>\> If you're using AI to code, you know the pain. Too many terminal pains, multiple agents running at once, copy-pasting context between them, and trying to remember which branch has what. The bottleneck isn't writing code anymore, it's coordinating agents. Intent is a developer workspace built for orchestrating agents, not just running them side-by-side. It starts with a living spec that updates as agents make progress. So every task stays aligned with no manual coordination. Intent works best with Augment's Augie and their context engine, but you can also bring Claude, Codex, or Open Code. Intent is what comes after your IDE. Try it yourself at augmentcode.com/intent. That's augmentcode.com/intent. Build with intent. And now, back to the episode. \>\> I think our view is like there's two good companies to start now. There's the AI native company that pushes the ball forward inside of some category, or there's the AI durable company that effectively uses AI where the core of the machine is not going to change. And if you look in our first two you know like there's no such thing as an AI native crypto currency. You know it's just like it's \>\> Yes. \>\> dramatically changed. Yeah, exactly. Like \>\> Put it on the blockchain and use AI and then uh \>\> The uh and and you know like we're not expecting it a robotic injector anytime in the next 7 to 10 years. And and so the the like core work of a med spa will be uh you know med spas themselves are actually like conveyors of the latest technology, the latest medical technology. Um you know when GLP-1s come out like med spas are one of the early adopters of of actually spreading it to their communities. Um and uh but at the end of the day like their what happens inside the walls of a med spa is not deeply impacted by AI. Um however, there are like around the edges spots where it can really matter in terms of reaching the right customers, serving them and communicating with them effectively at all hours of the day at an affordable price for the business. And so we want to be great deployers of AI inside of our operation. We want to help our partners you know deploy it to the maximum effective degree. You know we want them to be on the early early edge, not the bleeding edge necessarily. There's like no no need for them to be there. Um they can always just be a few months later and you know not take risk with their customer relationships. Um and um and so I think we've we've kind of said for now between the AI AI forward and the AI durable we really like being like great users of it inside that AI durable category. Um and maybe we'll end up changing our minds and finding the other category is really fun, but right now it's like uh every idea we've had in that category has had multiple formidable looking competitors doing something almost exactly like our idea there. And we tend to want to look at a category and say, we've got something interesting in our idea to say it's like worth our time and feels exciting to us. \>\> All the smart people are listening to Dan and Avery, and we'd rather compete against the less smart people who \>\> rather not compete against you, so I I love that. \>\> \[laughter\] \>\> So, if you're if you're thinking about, okay, how do I make this Well, how am I concentrating on the things that are not going to change and how am I using it to get AI to get more operationally efficient rather than being like truly like totally AI native in like, you know, a way that a YC company would be or whatever. Um what are the like actual real operational efficiencies you you have found? What's worked and like what has not worked as well as you thought it would? \>\> Maybe I could speak to the like company discovery process where I think we've actually seen probably the greatest transformation, which which roughly like maps to the the more greenfields, the better AI can be, basically. And like, whenever I talk to my founder friends that are seed stage, they're like, "Oh my god, our engineering is 10x faster." And then I talk to the like, you know, series D friends, and they're like, "We're like 10% faster. Like, what is everyone talking about?" Uh which is the sort of like classic dichotomy right now. The um in the in the new company discovery process, we are roughly speaking, like every stage has been rethought. Um so, the the first step is like, let's find some verticals to go start to poke around in. And this used to be like a week of Googling and maybe like calling some friends to just get the basic facts. And this is now like a mega prompt to like generate a list of categories, and a mega prompt to assess like what we think is a good business, and like our particular point of view of what we're looking for, and start to narrow in on a couple different um a couple different ones to go talk to real people in. Um and we did a we did a little bit of an exercise this time around where um we both did the like AI curation and then I did like a a human point of view and I actually felt in that moment like do you remember the children story of the it's like a myth of the the guy who was competing against the like automated tunnel builder like machine to build a tunnel? \>\> That's a real thing. Like John Henry or something like \>\> John Henry, yeah. Yeah. Well, it's a real thing. It's not like a guy put a hammer through a mountain. \>\> Well, he he I think did actually try to compete against the the automated thing and then died on it. \>\> Yeah, I'm pretty sure. \>\> Yeah. \>\> \[laughter\] \>\> All right. All right. RIP John Henry. Anyways, I was the John Henry in that story and uh fortunately came out the other end um and it and we did as a group select three categories. One of the three was like my human point of view. Two of the three were like the AI was like the No, no, no, these are screaming matches. And um and then more interestingly, we uh we built a insight with this AI ultimately forced us to move from Google Docs to Notion, which I was like fighting for years. And um we built an agent identity that we call Matthew Bolton, which is like a horrible name cuz Matthew Bolton was the Bolton or Watt. And uh Matthew Bolton is our assistant in being really good in the customer discovery process. So, he helps us prepare for every call um and like looks at the persona of who we're talking to, looks at our current hypotheses and and what our like validation focus is, and basically says like here's the areas to dive into. And of course, we review and make sure we like talk about the right things, but um he he makes the prep more efficient and then afterwards, the like AI transcript goes directly into a Notion table. We run Matthew Bolton and it regenerates a a point of view on for each of our core hypotheses on the idea. What's totally validated, what's totally invalidated, where do we need to dive in more? It pulls out the the relevant quotes from the people. So, it's been a huge hit in that way. Where it's totally missed for us, which is also kind of interesting, we've talked to other people who do these like synthetic customer calls effectively. They like make AI into the customer. And we just can't make that. We haven't yet been able to make that work at all. The the Basically, anything that strikes us as a good idea, so like passes some basic sniff test, the AI is like, "I'd love to buy this from you. Like, I'm so You know matter what we do, it's like it's like, you know, it it just expresses a a 10 out of 10 customer poll. And and so like we we try that and then flew to meet a real prospective customer for a category and just like fell on our face. Completely. And you know, have iterated our way past that, but like we just kind of don't think it's actually useful for that. It doesn't seem to um it doesn't know the nuances of the psychology of the like I've worked in this industry for 15 years and I'm deciding what to buy right now. \>\> And or it's like just it knows that you want it to say yes. So, it thinks you want it to say yes, so it does kind of \>\> We desperately tried to have it be like in sincerity mode and we can't get it out of that. I mean, so maybe someone else can. There's like a a way to make that useful, but I would say at the moment our point of view is like it can help us talk to people effectively, but it cannot actually reduce the number of people we need to talk to to get to confidence. \>\> Got it. Okay, this is really interesting. Are you able to show us Matthew Bolton? Like, show us the show us the goods. And I'm also curious about this prompt the prompts that you're using to do like, "Okay, research business ideas and also like filter them. That sounds really interesting, too. \>\> Okay, so we we have different Matthew Bolton flavors. I think there is an underlying like agent identity doc. This is for This was for a specific category dive we were doing into P&C insurance, and we kind of break down our own thinking across a few notes across a few docs. So, one is we have what we call a POV, a point of view on um what we think the opportunity is, what's the problem we're solving, who are we serving, um what what what \>\> Uh P&C is is property and casualty insurance. \>\> Okay, got it. So, that's the that's the area you're looking at. \>\> That's the area we're looking at, exactly. So, this is the the vertical we're exploring. So, we have our we have our point of view, we have a hypothesis tracker with a list of effectively What do we What are the core things we believe uh about the category, what has to ultimately prove true, what are we looking to discover through potential customer calls, and um and then we have transcripts and our own notes from every call we've had. And so, when we when we call Matthew Bolton, we ask it to reread our latest point of view, we ask it to read the hypothesis tracker, read the most recent 10 calls, and then for every hypothesis just re- just re-update effectively what's the evidence for that hypothesis, what's the evidence against, what's the strength of it, um and, you know, basically point us in the right direction. Tell us like where to spend more time, where to say like great, we're good to go on this. Uh \>\> This is kind of sick. I love this. \>\> Oh, thanks. I feel like that's actually really meaningful. You know, I got to say Sam and I this morning looked at each other and we were like, are we sure we should be on this podcast? \>\> \[laughter\] \>\> I literally said to Dan, I was like, let's make a list of all the places we've tried to use AI and it hasn't worked cuz I think yeah. Um so it's nice to hear that from you. \>\> \[snorts\] \>\> Yeah, exactly, exactly. We're like from the guy himself. Um the uh So you you know, ultimately you just say like run PNC analysis and it runs through all these different steps and um one of what what feels like personally meaningful about this is we try to be intellectually honest with ourselves. This is like something we we hold ourselves to and I think you probably know this cuz you're building new products all the time, but like the nature of starting something new is it requires like a manic energy and a little bit of a suspension of disbelief because there's just like no reason any new company should succeed or any new product should succeed and this keeps us like really rigorous and fact-based, um which is what we aspire to and and we can balance that with like our own you know, we can be sick offense to ourselves and ask it to remain fact-based and balance these two opinions. But it it like totally helps. \>\> Do you find that it's it works with the like cuz I find if you're if you ask it for reasons against or for, it's going to come up with it can come up with anything. So like do you find that it is actually good at weighing evidence for you or are you you're just surfacing it and then like making your own conclusion based on the evidence that it surfaces for and against what you what you believe? \>\> If we try to ask it an opinion on a high-level question, it we I I have not like I have not given I've not trusted it with that. Um or I you know, I I'm tend to take those results with skepticism, but I think it's really good at like finding the quotes. Um you know, at the end of the day like to support a hypothesis, in a perfect world what I want to bring to Sam is here's the three key things that must be true in this idea, and I've got three quotes from different people that like directly speak to each. And this will just like much, much more efficiently help me get there. Um and you know, maybe or maybe not I'm you know, we we ultimately meet that bar or maybe we lower the bar cuz we just find there's like something core and exciting in an area, but um the this keeps us honest um on the detailed things we ask you to do. \>\> And And somebody that's the role like Dan and I play for each other like is keep each other honest. And there's a way to sort of like get the short version of that with AI, too. Where you're like, okay, make the best counter argument. Um and and then it sort of just can like sharpen your thinking even if you're not looking for it to like tell you you're wrong, but you can sort of help it have it tease out um like what In what ways could you be wrong? Uh which is a really useful um tool. \>\> Sometimes I say that and I'll and then it you know, it makes some good counter argument and I'll be like, "Oh, off. Like, what are you saying?" \>\> \[laughter\] \>\> We like to move really fast at Every. We're always shipping new features and we run five products internally \[music\] with just 20 people. It used to be that updating any of our landing pages was a huge time suck. But then we started to use Framer. Framer is a website builder that works like a design tool, which means that the people who are closest to your brand and your customers can shape your landing pages themselves. Non-coders on our team use it all the time to ship website updates, and the results honestly are beautiful. It's really easy to use. There's real-time collaboration, which means your team can design together, see changes instantly, and ship faster. A robust CMS gives you everything you need for great SEO and content management. It also has advanced analytics with integrated AB testing that lets you optimize and improve continuously. Then it's just one click to publish. Your changes go live in seconds, so you can iterate at the speed of your ideas. And it's not just us. Companies like Perplexity, Miro, and Mixpanel trust Framer for their websites because the tool is enterprise-grade. They offer premium hosting, enterprise security, and 99.99% uptime SLAs. Whether you're launching a new website, testing landing pages, or building out your full site, Framer gives you the tools and support to move fast and build something great. \>\> \[music\] \>\> And everyone on your team can use it. Learn how you can get more out of your dot com from a Framer specialist or get started building today at framer.com/dan for 30% off a Framer Pro annual plan. That's framer.com/dan for 30% off. Rules and restrictions may apply. Seriously, we use this tool all the time. You should really check it out. \[music\] And now, back to the episode. \>\> And and when you're talking about like feeling outmoded on the research front, is that at what part of the research process? Because I also feel like if I was going to send you off versus Claude off to do some research, I am sure that Claude would cover like a wider breadth, but I feel pretty confident that if you gave me a research report on a sector, like I would like it better than the one that Claude made. Um it would just take you a lot longer, but I'm curious. \>\> I'm not that confident job. Yeah. I think it will do like a quite good job on basic market industry reports. I mean, it's it certainly beats for the one week desk research up front. Um it beats the like pretty low and medium quality reporting that was out there. There's there's kind of good stuff in We we really like public company filings whenever there's public company as in the in the category that we're looking at. And sometimes we're not just asking it to go do something. We're like uploading a bunch of thick PDFs, too. Um and it's a quite a good parser of that stuff. \>\> That makes sense. I think maybe what I'm saying is I would trust it to give me the consensus opinion about how people think about a particular space. And I would trust you to even if you were using AI to do it to come back with something that felt new and interesting in a way that I don't think that Claude can get to on its own. \>\> I I that I probably agree with. Like I I don't think we've gotten good ideas from it. I think we've I think we've uncovered facts and then we have our own earned point of view. Like we learned a lot through building Moxie and for 2 years have been looking for another category where a Moxie-style business might be a good idea. Um and we have a relatively differentiated point of view on actually what Moxie is and what makes it successful. Um and we have talked to between Sam and I probably 60 people starting \{quote\} business in a box in different categories, most of which don't have the properties that we think are essential for that. And um and so there's a combination of the like earned point of view through years of building with its ability to like of course consume massive amounts of information and then like fit that to our point of view. \>\> Yeah, I mean and I think if you like ask Claude for example like give you a bunch of business in a box ideas, they probably wouldn't be ones that felt good to us. Because we have like certain analogies and ways of thinking and looking at it that we you can ask it to do the research and figure out the underlying like category properties. Um but like the default of that isn't the answer we want. \>\> Hm. And you guys have done a lot of aside from using AI in your sort of company discovery process, you've been doing a a of thinking about how to bolt on AI to your existing businesses that are not necessarily like AI native and doing the the transformation process there. What have you learned from that? \>\> Yeah, I think it's really interesting like inception companies, series A company, series C. And I think there are like two parts of this of like one, how do you actually get people to start using AI, which we talked about, and then then I think the second piece is like what's actually worked and what hasn't. Um Dan and I were talking about this the other day of like should you have an initiative, like an AI initiative? Is that a good idea or bad idea? And I think my perspective was like it's a bad idea because you don't want to sort of lead with the the hammer. And you know, there we all remember the time like no sequel, everyone was putting everything in no sequel, whether it belonged or not. Everyone was building a slack bot, you know, 10 years ago, whether one needed or not. But you like you do need something to kind of like shock the system. You need something to be like, okay, great, there's like a new tool set. And I think like the point of view I've come to is you shouldn't give anyone credit for using AI, but you should make sure that the expectation is that they use AI, or sorry, the expectation is they'll deliver the best product and output knowing that AI exists. And and so to do that effectively, you both need to like sort of seed what are the tools you can use and give a lot of good examples. And you need to start like demanding that when you see the results from someone on your team that they've actually used the best possibilities. But you don't get any points for generating a bunch of copy that's clearly written by AI and is bad to read, right? Like you have to vet the copy, but like if you use a prompt to get 70% of the way way there or even 100%, that that's great. Um and so like that's where you found it, like you sort of have to like, at least I found it, um you sort of have to figure out who are the people in your team to seed these ideas so other people get examples and then actually make it successful for for those folks. I think like one of the places that um and I think this goes to Dan's point earlier of like where we're seeing the most action in inception is really like on the experimental experimental edge of things. So, um a few examples of this like really good at generating landing pages and like pushing our thinking there and coming up with stuff. A ton of work to like integrate that back into Webflow and have it fit with our system and have a consistent header and footer. Um and so there's almost like two phases. There's like research development and then there's like production. Um we've had um a number of people on the team build sort of like throwaway um apps um that've been really useful. So, for example, one of the challenges we have is for the few funeral business people call in and they might mention some town name and we know we know whether we service that. So, a designer spun up an app where anyone can type in a hospital name, a town name, and it resolves whether we can find it or not. Um instead of like integrating that and spending engineering time on like figuring out how to deploy that safely, integrate it, all that stuff, it's just a separate app that's like a link um from our main one. Um and so we found that like enabling those types of things has worked really well. Um and our engineers have gotten more productive, but a lot of the core things an engineer does is still the same, right? Like they're faster coding, but deployment still takes work, maintenance still takes work. And so, it's almost like we have sort of like two different pieces that are being enabled by AI, but like in parallel paths um today. Um \>\> It sounds like you have like uh the greenfield things are quite fast and then anytime it has to touch something that already exists, it's like a speed up, but it's not totally changing everything if that makes sense. \>\> Yeah, I think that's exactly right. Um and there's some places where it's like made a huge difference. I um our town team was reminding me uh the other day that they've made something called Sam GBT, which they trained on all my blog posts and they use to reach out on my behalf to potential candidates on LinkedIn. And so it's like trained my voice, I forgot it existed, and that's worked really well for them. And so there's sort of these like places where it kind of come unlock and enable into existing system. \>\> I also think it's interesting kind of the approach of you don't get credit for using it, but you do get credit if you basically you have to just do the best possible thing given that AI exists. The thing that that seems to solve is what you don't want is people just doing Potemkin villages of like I used AI for all this stuff and it's just like it's all just for show basically. Um, I'm curious though because my experience we do a lot of like AI transformation type stuff with big companies and my experience is if you don't if you just say that then people will just continue to doing it the way they already know how to how to do cuz they're like well that's if I want to get the best quality I have to do the thing I already know. So how do you deal with that? \>\> I think it's like finding a few people to set the example and then start comparing the work. And so like we don't count like oh how many PR like how much of your PR was AI generated because like if you commit and push a bad PR because AI wrote it and you're like blame AI like that's a terrible outcome. Instead we can be like oh look at this engineer who's like done all this great work. How ask them how you do it in a public forum and then sort of like continue to like raise the the floor. Um, and so that's what we found to be much more successful and it takes a lot of active work to say okay, let's find these examples across the company, let's potentially seed those examples, and then continue to elevate elevate those. \>\> Have you guys seen any change since Opus 4.5 came out? I feel like there's been a big change at least for us and just on the broader like X sphere has that filtered into any of your businesses since they sort of exist in a in a slightly probably less tech forward part of the economy and part of the world or is it are you still kind of in the regular chat GPT type wave? \>\> The the early word out of the Max engineering is like yeah, this is better but not this is a step function different experience and we are doing a lot of work there to like retool in order to experience more benefit because we're seeing exactly they're like newer engineer working on more green field project moving much faster than you know, working on something that like touches multiple parts of the system. Um and that's like, you know, 40 person product engineering team more mature code base and so on. Um so we have not seen the like night and day transformation that the X sphere is reporting. \>\> I I would say I I we saw sort of like similar phenomenon like it feels like over the last year where everyone's talking about like Geo instead of SEO and like everything's going to change and generative commerce. And I and I think like that's one place also where we've seen much more incremental change. We're getting more traffic from chat GPT um but it's almost like it's another channel for us and we have to think about the same way we do like paid search. Um there's like a cat mouse game to figure out how to get free results. There's going to be a paid version of it. But fundamentally like one, we don't think people wouldn't buy a funeral um via chat. Um and two, um that may not be true for a lot of products, right? Like when I do a flight search, I still prefer to do that myself um versus like ask a travel agent to do that. And so for us it wasn't really a shift in the way we thought about marketing. It was really like, okay, great. We have to pay attention to like these five channels. Here's another one um that we have to make sure we're ahead of but also doesn't actually change our business fundamentally. \>\> Can you guys give me a preview of what your next business might be or what areas you're interested in? \>\> I think we cannot and and I don't know if Sam knew I would say that but uh I think we cannot not because we're like hiding a secret so much as the level of embarrassment at like literal day zero is so extreme that I do not think I can tolerate it. I think we just made we're not like we're not like 30 days from launch and announcement. Um we're we're a few months out. We we do know now what it is um but that's literally as of the last like five days and so we're in that like we we just have to build more before we feel a little bit more comfortable um talking about it publicly. \>\> I I think I mean I think we can maybe go back to the original theme. I think like we I think it's like a useful distinction to be like okay what are like things that are like purely enabled the unit of work has changed um and then there's sort of this other category of like it's some combination of like hard bundle in the real world um and those types of things we're really interested in. Um so like the P&C insurance one um that that Dan listed there's some transformation aspects to it and there's some things that are like kind of going to work the same. Um and so I think we're looking for some sort of like secular change to the world like the world is changing um and it could be you know the death rate's going up. It could be that more people want to do Botox. Um AI is sort of this mega trend but we're sort of like looking for the intersection of AI in some other trends rather than AI being the primary trend. Um and so that's sort of like helps guide the types of things we're we're interested in. \>\> When do you start to talk about new your new products, Dan? \>\> I mean, I've been like my all my products start as basically blog posts of like I built this little thing over the weekend. Like, you know, uh so, pretty much immediately. I think of new product uh new products as content first um and then in businesses second if they seem like they have legs. \>\> Mhm. That's cool. \>\> Yeah, and it's been like so fun to watch you share all these like little toys and experiments because it feels like um this sort of like edge of what's fun and cool uh and like potentially useful. But, like you almost like don't fully ask the last question, which actually I think I really appreciate because it takes you into like all these new places. And and our business are like, "Well, will someone pay for this?" Uh you know, like it it's sort of like is the first question we ask. Um and so, we're both doing something kind of weird, but like in really different directions. \>\> Totally. Totally. And what I think is also fun is I don't know if you remember this, but when I started Every, you were also thinking about newsletters and you were thinking about newsletters in a almost similar kind of way, but it it was very it was very you. It was like I think you you were doing cars at first. You were running a vertical specific newsletters that like people would pay for. So, like cars or trucks or whatever. And yeah, so there's just I don't know. We're always kind of like on parallel tracks, but uh with very different personalities that come out in the way that uh the businesses actually get built. \>\> Totally. Yeah. Yeah, I think we were running a um automotive professional publication. Um and the idea was that I do like a vertical stack version of information, uh which ended up not working for or maybe executed a different way would have worked, but like it it's like yeah, the we're working for professional uh newsletter and then they're two completely different things. \>\> It's yeah, it's very hard. \>\> This was pre-built in a week. I just want to put that out there. \>\> That's pre-built in a month. Yeah, Dan doesn't want to be associated with uh \>\> You didn't You wouldn't have done cars car news letters. \>\> \[laughter\] \>\> Yeah. \>\> \[gasps\] \>\> Um cool guys, this is awesome. I love having you uh whenever your new thing launches, I would love to have you back to talk about uh how AI is is uh is involved in that and um yeah, just just love being on our on our parallel paths in uh in in Brooklyn together. \>\> Next time in person. \>\> Sounds good. \>\> Oh my gosh, folks, you absolutely positively have to smash that like button and subscribe to AI and I. Why? Because this \[snorts\] show is the epitome of awesomeness. It's like finding a treasure chest in your backyard, but instead of gold, it's filled with pure unadulterated knowledge bombs about Chat GPT. Every episode is a roller coaster of emotions, insights, and laughter that will leave you on the edge of your seat craving for more. It's not just a show, it's a journey into the future with Dan Shipper as the captain of the spaceship. So, do yourself a favor, hit like, smash subscribe, and strap in for the ride of your life. And now, without any further ado, let me just say Dan, I'm absolutely hopelessly in love with you.