An interview episode with Ben Tossell, founder of Ben's Bites and former Makerpad founder (acquired by Zapier), on running a profitable one-person internet business using AI. Topics include using ChatGPT as a business strategist, building an MVP with ChatGPT, generating landing page copy, extracting information from research into articles, refining writing with Lex, analyzing business and customer feedback data with AI tools like Claude, Gemini, and Julius, and offloading time-consuming tasks to AI. Timestamps and links to resources mentioned are included. # Video Identity Observations — Ep. 14 with Ben Tossell ## Video Title **"How to Run a Profitable One-person Internet Business Using AI - Ep. 14 with Ben Tossell"** - The title is **episode 14** ("Ep. 14") of an **interview/guest-show format** series, indicating an ongoing episodic podcast or video show. - Title frames the channel's claimed scope: **using AI (specifically ChatGPT and adjacent tools) to build and run small internet businesses**. ## Explicit Self-Description Evidence ### Host's stated show concept The host describes the show's premise directly: > "I think that's definitely the concept of this show is there's so much um latent hype or latent energy around ChatGPT. Like people just feel like they should be using it, but not enough people actually know how um because it's such a general tool..." He further summarizes: > "Giving people the actual the real ways that that you use it or I use it or or lots of other smart people use it is the most valuable thing you can do at this point in the AI wave." ### Show name stated in the transcript The closing segment names the show: > "smash that like button and subscribe to How to Use ChatGPT" - **Channel/show name: "How to Use ChatGPT"**, hosted by **Dan Shipper** ("a journey into the future with Dan Shipper as the captain of the spaceship"). - Format includes a **"show and tell portion"** where guests share screens and demonstrate workflows live: *"I would love to just get into the sort of show and tell portion. I know you have a bunch of stuff to show me about how you use chat GPT."* ### Host's own recurring framing The host presents a signature idea he has written about: > "I think starting to think of yourself as a model manager rather than an individual contributor is like a really powerful idea... it's one of the recurring ideas on this show" - The host also references **past episodes** ("this is mentioned in the episode you did with Staff" / "your episode with Stef Smith who did Internet Pipes"), demonstrating an established episode archive. ## Intended Audience - **People who want practical, real-world AI usage** rather than hype: *"we're less in the fancy demo hype phase and more in the like we're actually starting to use this."* - **Aspiring or current founders of small/bootstrapped internet businesses**, especially one-person operations: *"anyone who wants to start profitable internet businesses"*; *"for people who are watching or listening to this and haven't run a business themselves."* - **Non-technical or non-writer operators**: the guest notes "I'm not really a writer, which is sort of ironic," and the host frames tools as removing drudgery for solo operators. ## Claimed Scope of the Channel - Demonstrated scope: **long-form interview conversations with AI practitioners and founders**, including screen-shared walkthroughs of actual ChatGPT/Claude/Gemini/TypingMind/Lex/Julius workflows. - The host claims the show covers **genuinely useful AI applications by real people**, positioning against "latent hype." - The closing promo (clearly a stylized/satirical sign-off) claims the show delivers "pure, unadulterated knowledge bombs about ChatGPT" — a promotional claim, distinct from demonstrated content. ## Notable Presentation Patterns (Observed, Not Claimed) - **Guests are credentialed operators**: the host introduces Ben Tossell with specifics — runs the Ben's Bites daily AI newsletter, invests in AI companies, exited founder (sold MakerPad to Zapier), ex-Product Hunt community manager. - **Host has his own business context**: *"we have five employees"* and *"we incubated it"* (re: Lex), indicating the host speaks from operating experience. - **Interactive live experimentation**: host and guest test prompts together in real time ("Should we try that? Let's try it."), a demonstrated format feature. - **Dual distribution implied**: the host says "for people who are watching or listening to this," suggesting video and audio formats. - **Affinity with newsletter ecosystem**: guests and host reference each other's newsletters (Ben's Bites), suggesting an audience overlapping with AI newsletter readers. ## Distinguishing Claims from Demonstration | Type | Evidence | |---|---| | **Creator claim** | Show is about "the real ways that you use it"; "recurring idea" of model manager; subscribe pitch in outro | | **Demonstrated** | Screen-share walkthroughs of ChatGPT, TypingMind, Gemini, Lex, Julius; reverse-engineering CB Insights' Yardstick; survey/CSV analysis; Stripe/Beehiiv reconciliation | | **Creator claim (guest, about self)** | "I'm the only one who touches Pro"; "accidental content media founder's journey" | | **Demonstrated (guest)** | Detailed, verifiable workflow descriptions with specific tools, prompts, and outputs shown on screen | # Delivery Observation: "How to Run a Profitable One-person Internet Business Using AI – Ep. 14 with Ben Tossell" ## Format Overview This is a **two-speaker podcast interview** with a "show and tell" structure: - **Host (Dan Shipper)**: facilitates, frames themes, summarizes guest points back to them - **Guest (Ben Tossell)**: demonstrates workflows live via screen share, walking through ChatGPT/TypingMind/Gemini/Lex sessions Notably, the transcript ends with an **AI-generated outro read** — hyperbolic promotional language absent from the rest of the episode ("smash that like button," "I'm absolutely, hopelessly in love with you"). --- ## Host Delivery Patterns ### Reflective summarization ("one thing I'm seeing you do...") The host repeatedly restates the guest's behavior in an analytical frame: > "One thing I'm seeing you do is when you think about a company and company that makes a product, the thing that you can see on the internet is sort of like the tip of this iceberg." > "What I'm what I think this episode is about is what one thing you're really good at is starting these one-man profitable internet businesses that are like sneakily big." This summarizing move occurs at least three times across the episode — a stable habit, not an isolated moment. ### Vocabulary: frameworks and metaphors - **Iceberg metaphor**: "tip of this iceberg... underneath the water level" — used twice by the host. - **"Model manager" framing**: "starting to think of yourself as a model manager rather than an individual contributor" — a recurring show concept he names explicitly ("that's one of the recurring ideas on this show"). - **Drudgery**: "requires like a decent amount of drudgery" / "It doesn't feel like that drudgery anymore." ### Scaffolded questioning The host guides with sequences of short setup questions and confirmation checks: > "Is that Is that right?" > "Did that answer your question? Like did that Was that what you wanted?" ### Cooperative live prompting During the screen-share, the host co-writes prompts aloud, with visible hesitation and revision: > "Um, like, it's probably something like, um, in in the screenshots I'm attaching um, there are questions that, um, hm, let me think about it." ### A signature playful tip The host shares a prompting idiosyncrasy with hedged uncertainty: > "I'll give you $2,000. I'll tip you $2,000 if you do it right. I've never actually tried that... I don't know if it works, but I do it all the time." --- ## Guest Delivery Patterns ### Filled pauses and stutter-restarts The transcript is dense with repeated-word restarts, from both speakers but especially the guest: > "where where did that all come from" > "I I've I've learned some from from this podcast" > "It's It's effectively like a small snippet" This is consistent throughout — a stable speech feature, not transcript noise in one spot. ### Hedged, self-deprecating framing Tossell frequently qualifies his claims or undercuts them: > "which is not true. It's a Forta customer" > "could be completely hallucinated" > "I'm pretty sure I wasn't a good manager when I ran my company" > "I'm certainly not myself like a manager. Maybe it's a branding problem" ### Teaching habit: narrating the demo step-by-step During screen shares he reads what's on screen and narrates his exact prompts: > "So, I thought that's really interesting. I wonder if I could throw that into ChatGPT using image search and say like, 'What's in this image?' And that's literally how I started the conversation." > "I said, now look at the quotes on this image. They are snippets of other responses from different customers... Tweak the questions you just wrote based on what is mentioned in the image." ### Recurring vocabulary and phrases - **"Rabbit holes"** — used at least four times ("This actually sent me down another rabbit hole," "going down rabbit holes," "how does that get to Glassdoor and then how does that get to G2"). - **"Unfiltered"** — repeated when describing reviews ("crowdsourced anonymous unfiltered review. I put used unfiltered there"). - **"Simmer that down"** — "if I sort of simmer that down to a simplified version." - **"Fire hose"** — "we always joked about the it being just this massive fire hose." ### Explicit honesty about limitations A consistent habit of flagging what the AI did *not* do well: > "It kind of summarized it into one bullet point rather than two" > "it just gives you really generalized stuff" > "They didn't actually put in the quotes. So, I think that's fair enough." ### Self-aware speech commentary > "cuz I I kind of write how how I speak. So you can probably tell if you're listening to this that I kind of speak all over the place" --- ## Shared Communication Patterns - **Mutual mirroring**: both speakers repeatedly affirm in doubled form — "Yeah, yeah," "Totally. Totally," "I love that. I love that." - **Cross-reference habit**: both anchor points to prior episodes ("the episode you did with Steph," "Sahil showed..."). - **Enthusiasm markers**: "That's amazing," "That's awesome," "I love that" recur from the host as conversational punctuation. --- ## Uncertainty Notes - Timestamps are unavailable beyond rough transcript indices; pattern frequency claims above are based on counting occurrences in the transcript text only. - Filled-pause density may partly reflect transcription artifact rather than pure speech behavior; the pattern is too consistent to dismiss entirely, but exact frequency should be treated as approximate. - Speaker attribution for some lines (e.g., overlapping quick exchanges) is inferred from context, not labeled in the transcript. # Position Record: How to Run a Profitable One-person Internet Business Using AI — Ep. 14 with Ben Tossell ## Participants - **Ben Tossell** — founder of Ben's Bites (daily AI newsletter), Ben's Bites Pro; previously community manager at Product Hunt, founder of Makerpad (sold to Zapier); AI investor (identified from transcript and host's introduction). - **Dan Shipper** — host; frames the show's premise; mentions incubating Lex. ## Framing and Big-Picture Positions ### Dan Shipper (host) - **The AI wave has shifted phases**: from "fancy demo hype" (tweeting GPT-3 demos for instant likes) toward "we're actually starting to use this." - **"Model manager" framing**: thinking of yourself as a *manager* of AI rather than an *individual contributor* is a powerful idea — a good manager knows what the employee is good at and how to get the best out of them, which transfers to AI. - **Iceberg observation**: what's visible of a company online is "the tip of this iceberg"; studying it closely reveals "what's underneath the water level." - AI is a **leverage point** for small profitable internet businesses — reduces time/money cost of the many small operational tasks. - Closes with (unverified playfulness) a heavily promotional outro; no substantive position. ### Ben Tossell - **Hype fatigue in daily coverage**: by 2024, covering "chat with your PDF product number 57" is exhausting; the "massive fire hose" of AI news continues, and audience appetite may be shifting from "tell me a new tool every day" to "how do I actually use this." - **Acknowledge his own bias/jadedness**: he sees everything ("another dating app, another weather app" from Product Hunt days), while ordinary users may see one ad for one tool and find it genuinely useful — e.g., PDF AI reportedly making ~$200k MRR from a single niche. - **Disagreement (preserved)**: he pushes back on the "model manager" branding — *"I don't like being a manager. I don't want to be a manager... I'm pretty sure I wasn't a good manager when I ran my company."* He agrees the underlying skill (which tool, how to prompt) matters, but rejects the label. Shipper: *"Maybe it's a branding problem... I'm sticking with manager."* - **Default to AI**: he tries to use AI "for everything small and bigger... at every moment of the day," preferring "the lazy work of prompting" over clicking blue links — but notes even he forgets to apply tips/tricks multiple times a day. - **Business philosophy**: most people over-index on the VC/TechCrunch "shoot for the moon" model; he believes **"anyone can build most things"** into a multi-million-a-year lifestyle business. He's not seeking a billion-dollar outcome. He notes "lifestyle business" branding has somewhat "ruined it" even though a $2M/year lifestyle business sounds great. - **Against procrastination-by-information**: "No one needs any more information on how to start a business... you're just procrastinating on the fact that you could just do it today." Rabbit-hole research is done because it's interesting, not as a prerequisite. ## Demonstration 1: Researching Companies via ChatGPT (CB Insights' Yardstick) ### Tossell's method and claims - Yardstick (CB Insights spin-off, main product reportedly tens of thousands of dollars/year) sells anonymized buyer interview transcripts (role, purchase amount, structured sections: introductions, evaluation, sales experience, deployment, competitors, pricing, post-deployment). - His curiosity is threefold: 1. Fascination with **data products** — he believes a one-person company could build a niche database product reaching ~$500k/year. 2. Learning "what the best of the best are doing" and simmering it down to a simplified version for a different category. 3. "Peeking behind the scenes" of enterprise products normally gated behind sales calls. - **Workflow**: paste screenshot → "What's in this image?" → ask ChatGPT to guess the questions behind each transcript section → feed additional screenshots (7 transcript images plus an "Evendo Scorecard" showing satisfaction scores, win reasons, opportunity areas) → refine questions. - **Honest uncertainty (preserved)**: he can't verify the generated questions — "all these questions could be viable... could be completely hallucinated." He resolves this by gut instinct: does the question plausibly elicit the observed answers? ### Shipper's coaching during the demo - **Split the thinking step from the editing step**: ChatGPT often performs worse when asked to generate and edit simultaneously; have it first articulate what questions produced the answers, then modify. - Multiple images at once are fine; the generation-plus-compression step is what's "complicated for it." - He jokes about tipping prompts ("I'll tip you $2,000") — says he does it all the time, doesn't know if it works, "it's just fun." ### Outcome claimed by Tossell - This exercise seeded a real project: **anonymous AI-product reviews** (Cursor, LangChain, Copilot, ChatGPT), planned launch at requestforai.com. - Follow-up prompt suggested generalized steps (research, website, etc.) which he skipped: *"Focus on the content collection part. I know the rest."* - Reward model considered: contributors get access to all reviews ("give to get"). ## Demonstration 2: Glassdoor / G2 Rabbit Hole - Tossell researched Glassdoor's founding and monetization (job advertising, employer branding) via ChatGPT. - **Tension he flagged**: employer-branding monetization "doesn't feel like it fits the purpose of the site" — the reviewed company trying to entice job-seekers on the same page as unfiltered reviews. He personally signed up and confirmed the **give-to-get model** (contribute a review/salary info to unlock content). - **Critique of G2/Trustpilot-style monetization**: monetizing via leads to reviewed companies "feels like a really backwards way of thinking of that model," though he concedes the model clearly works (millions of monthly views for several companies). As a user, "you actually just want to know what people really think." - He links this exploration to Shipper's episode with Stef Smith on "Internet Pipes" — rabbit holes don't need to connect, but pieces may spark ideas. ## Demonstration 3: Running Ben's Bites Pro with AI (Writing Workflow) ### Tools and claims - **TypingMind**: ChatGPT-style interface with model switching (Claude, GPT-4, Gemini), no token limit, cost display — he's "happy to spend 14 cents" extracting info from a transcript. Previously used a free Claude "AI Influencers" account (says he was kicked off) and the Glasp Chrome extension for podcast transcripts. - **Gemini + Google Docs**: pasted a long transcript into a Doc and prompted Gemini to "Highlight key topics I should write about... Include quotes." Gemini returned topics (e.g., "Ignite Tech using AI to drive cultural change") with real quotes, some of which became sections in the final post. - **Claim with qualification**: "It doesn't write the blog post for me. So I'm not claiming that it does that, and nor would I want it to." AI is "this strange writing partner" — 6 minutes of setup versus 6 hours of manual quote-pulling. - **Lex (lex.page)**: used for drafting with versions — he stores AI drafts from Claude, GPT-4, and Gemini side by side, then writes his own version 2. Rationale: going from "something really horrible" to his own words is easier than a blank screen; parallels remixing proven business models. - **AI "run checks"** in Lex: used only for gravity and readability; he selectively ignores suggestions (e.g., keeps "radically changing their company," keeps British phrasing) — a stated example of human judgment over automation. - **Why he keeps it in-house**: "I want to be the one that's written this post. I don't want... an editor... none of my thinking has gone into creating that structure." - Claims he only launched Pro (one long post per week) because AI made it feasible; he's the only one who touches Pro. ### Host's synthesis - These tools let a one- or two-person business produce output faster and at higher quality — key leverage for profitable internet businesses. ## Demonstration 4: Data Analysis for Business Operations ### Tossell - **Subscriber survey analysis (ChatGPT + CSV)**: prompted it to summarize why subscribers stay and to run sentiment analysis. Noted it **did not follow the "ignore outliers" instruction** — it surfaced "inertia" as a stay reason (which both speakers joked about). He acknowledged he could have followed up on the sentiment stats (e.g., minimum −0.7) but didn't; it was "just a pulse check." - **Julius (investor)**: used to reconcile a Stripe/Beehiiv subscriber mismatch — identified 27 emails not in both files and which file held each, turning a potential "2-hour task" into a smaller one; cause was email changes not syncing across platforms, verifying payers keep access. ## Unresolved Tensions and Qualifications (Summary) - **"Model manager" label**: Shipper endorses it; Tossell rejects the framing while accepting the practice — disagreement left standing. - **Hallucination risk**: generated interview questions are unverifiable; Tossell's mitigation is plausibility judgment, not verification. - **Tool output quality varies**: ChatGPT sometimes degraded detail when merging questions; didn't follow outlier-exclusion instructions; sentiment follow-ups unexplored. - **Incentive-structure critiques** (Glassdoor employer branding, G2 lead-gen) are Tossell's stated reactions, not settled conclusions — he notes these models "clearly work" commercially despite feeling backwards to him. - Stated plans (requestforai.com launch "by the time the podcast comes out") are intentions, not confirmed outcomes. # Speaker & Provenance Record — Ep. 14 with Ben Tossell ## Video - **Title:** How to Run a Profitable One-person Internet Business Using AI – Ep. 14 with Ben Tossell - **ID:** 01a0f15b-30f5-7018-bae6-4704bd06c8c1 ## Speakers & Roles | Speaker | Role | Evidence | |---|---|---| | **Dan Shipper** | **Host / creator** | Welcomes guest, runs the show ("welcome to the show"), references "the concept of this show," states "we have five employees," and is named in the closing promotional segment ("Dan Shipper as the captain"). States his team incubated Lex. | | **Ben Tossell** | **Guest** | Interviewee; runs Ben's Bites newsletter, exited founder (sold company to Zapier), angel investor in AI companies. | ### Secondary/quoted persons (mentioned, not speakers) - **Eric Von Bohn** — quoted from an interview transcript (Ignite Tech CEO), discussed in the Ben's Bites Pro workflow segment. - **Sahil** — referenced as a past podcast guest ("Sahil showed just regenerating a bunch of stuff"). - **Stef Smith** — referenced past episode guest ("Internet Pipes"). - **Ryan Hoover** — referenced re: Product Hunt commenting tactic. - **Sundar (Google)** — referenced re: posting Google AI announcements. ## Referenced Sources & Third-Party Products - **Companies/products analyzed by guest:** CB Insights, Yardstick, Glassdoor, G2, Trustpilot, Product Hunt, Zapier, Snowflake, Teradata, Forta/Forter, Stripe, checkout.com, Yotpo, PDF AI, Cursor, LangChain, Copilot. - **AI tools used/demonstrated by guest:** ChatGPT, TypingMind, Claude, Gemini, Perplexity, Supernormal, Glasp, Lex (lex.page), Julius, Apollo. - **Guest's own properties:** Ben's Bites, Ben's Bites Pro, MakerPad, requestforai.com (stated: "will be live by the time the podcast comes out"). - **Ben's Bites Pro interview subject shown:** Ignite Tech CEO Eric Von Bohn transcript. ## Third-Party Claims (unverified, attributed to speaker) - CB Insights product "tens of thousands of dollars a year" (guest). - CB Insights revenue discussed "on podcasts" (guest, unsourced). - PDF AI making "~200k MRR" (guest, unsourced). - G2/Trustpilot-type sites getting "3–5 million views a month" (guest, unsourced). - Guest took "50 plus ideas" to reach MakerPad (guest). - Snowflake quote from Yardstick landing page (~70% cost savings vs. Teradata) — sourced to Yardstick screenshot shown. ## Promotional Segments (flagged) - **Closing outro:** Extended like/subscribe promotional segment for "How to Use ChatGPT" show ("smash that like button and subscribe… Dan, I'm absolutely, hopelessly in love with you"). Speaker unattributed; reads as a produced/promo voice, not Dan or Ben in conversation. - **Guest self-promotion:** Ben's Bites and Ben's Bites Pro discussed extensively; requestforai.com URL promoted. - **Host-affiliated product:** Lex is demonstrated; host discloses "we incubated it." Guest calls his Lex mention "a real authentic moment" — flag as incidental promotion with disclosed affiliation. ## Availability & Limitations - **No curated metadata supplied**; roles inferred from transcript only. - Guest could not access paid Yardstick product content ("I can't pay the thousands of dollars to go and sign up") — analysis based on landing page screenshots and CEO tweets. - Guest notes ChatGPT outputs may be "completely hallucinated" in the reverse-engineering exercise; he verified Glassdoor's "give-to-get" model by personally signing up. - Speaker labels for the two conversational voices are inferred from context (Dan's host lines, Ben's guest lines); no explicit self-introductions of "Dan" by name occur in dialogue outside the promo outro. - **Comment availability:** not stated in source material. One thing you're really good at is starting these one-man profitable internet businesses that are like sneakily big. The Glassdoor was is basically an anonymous site for people to leave a review on your employer. So, I was like, "Okay, well, where where did that all come from? How did that start? How did they start getting people to write reviews on a site like that? How did Glassdoor get reviews on the site initially? Did they run any any incentives?" So, I was looking at that kind of thing. So, things like that was like, "Yes, yeah, it's just an interesting way to like start exploring an idea and where it's going to go and how it how it changes over time. One thing I'm seeing you do is when you think about a company and company that makes a product, the thing that you can see on the internet is sort of like the tip of this iceberg. If you study the tip pretty closely, you can tell a lot of things about what's underneath the water level that you might not be able to see. And welcome to the show. Thank you for having me. It's great to have you. Uh, we've been friends for a while. For people who don't know, you run the Ben's Bites newsletter, which is a daily AI newsletter that I read every day. I love I think it's one of the best, if not the best, daily AI newsletters in the space. Um, you also invest in AI companies and you are an exited founder. So, you previously sold a company to Zapier. Um, so, you have you have a ton of experience as a founder and I think you're also just great at the content media thing and it's really fun to have you on the show. I appreciate the kind words and yeah, it was kind of an accidental content media founder's journey, but yeah, um, that's what I am. Yeah, I just sort of remember I think this is like a year ago, but I just started seeing this guy anytime someone would like launch an AI demo, there's this guy that would just like pop up underneath any one of their uh, any \[snorts\] one of their uh, projects being like, "Hey, I just added this to like my Ben's Bites newsletter. You should subscribe." And um, I was like, "Who is that guy?" Yeah, that was That was like an unintentional thing that then other AI newsletters started copying as well. And I I didn't realize it at the time. I was like, "Actually, that's what Ryan Hoover did at Product Hunt." And I used to work at Product Hunt before I went and launched MakerPad. Um He was just like, "Hey, yeah, I'll put this You should post this on Product Hunt." And then it must have been in my brain somewhere. And I thought I then started realizing that if Sundar from Google is posting something about the Google AI thing, if I could be quick enough to just put my link underneath, it was actually a really good way to start like getting subscribers and you're reaching the right audience in the right place. Um So, that was a a little hack that I didn't intentionally mean to be a hack, but it worked out well. Then everyone else started doing it. I stopped doing it. But, yeah. Um that's amazing thing. Yeah, I'm curious like just high-level I I feel like a year ago was this big hype wave for AI and I I think you and me both like rode that to a certain extent and like got a lot of traffic and a lot of um I don't know, just a lot of good stuff um out of that wave and I think that that has started to dissipate or at least it's changed um and we're starting to enter the like it's not just AI any It's not just like you can you can tweet uh like some demo that you did with GPT-3 and or just get a thousand likes like immediately. Um we're we're less in the fancy demo hype phase and more in the like we're actually starting to use this. Like I'm curious what how things have changed for you or what you're thinking about right now in this current era of AI. Yeah, I mean, on the business side, as a daily AI newsletter, I mean, we I started it because there weren't any others. And at the time, this was like October 22, it was pretty chat GPT, so it was definitely early. Um and then it was just riding that that wave that everyone else was on and seeing anything and covering everything. I want to know everything that's launched and all the new stuff every single day. And then as we've come to sort of now 2024, even my thinking of this is exhausting to get it out there. Um it's exhausting for someone who does this as a business to keep up with what's the appetite really like for everyone else consuming this stuff as well, where like chat with your PDF product number 57 is a new thing that on product hunt or or hacker news or whatever, like are we covering that again? Are we doing that kind of thing again? So, there's a lot of the same stuff. So, we yeah, we've we've sort of think thinking about that internally and yeah, we launched Ben's Bites Pro in December, where I speak to sort of similar actually to what you're doing, I guess, where I chat to businesses and CEOs and things of how they are using AI in their business and like changing their organization and rolling out new products and stuff cuz I just think there's not that much out there. There's no like playbook on companies building with AI. And I think it's really interesting, so yeah, we started doing that and getting some really interesting conversations, but yeah, I don't know don't know what the future holds with all of the daily stuff. Like there's just we always joked about the it being just this massive fire hose. Um it still is and like are people's appetite changing towards I actually want to use how know how to use this thing. I don't want you to just tell me a new tool every single day. Um I I feel that. I mean, I just love how bored you sounded talking about the the the next PDF tool number 57. Well, I I mean, again, this all still goes back to my Product Hunt days. I was I sort of somehow made I was a community manager. It was my job to like let companies be on the homepage and like support their launches. And also like get in the comments and start people like talking about it. And then I got to a point where I must have just It was probably a similar time frame where I was like, "This is another dating app. This is another weather app. Why do we need this? What's the What's the point of this? There are so many others." Um and maybe I haven't changed all that much since then, I guess. Um But I think it's just I like seeing new things. I like seeing things that are different or there's a reason behind doing this other than "Oh, I've just done the same kind of thing as everyone else, but it's like slightly different because of this one thing." But equally you see all these PDF products, there's plenty of them. You can do it in ChatGPT. And then there's companies like PDF AI making, I don't know, 200k MRR from just that focusing on that one thing. So, it's definitely like a bias between I'm seeing all of this stuff, other people are seeing maybe that one ad for that one tool and signing up and using it and finding it useful. Um So, yeah, that's how I'm That's why I'm biased and like jaded by it, I guess. No, that makes perfect sense. I mean, I think we're both sort of in pursuit of like what is genuinely and I think we both have moved from any any new AI demo is interesting into like how are people actually using it? And I think that's that's definitely the concept of this show is there's so much um latent hype or latent energy around ChatGPT. Like people just feel like they should be using it, but not enough people actually know how um because it's such a general tool and there's so many different ways you can use it and even just remembering to use it is like kind of hard. And so, yeah, I feel I feel like we're all we're sort of on the on the same wavelength there. Like giving people the actual the real ways that that you use it or I use it or or lots of other smart people use it is the most valuable thing you can do at this point in the AI wave. Yeah, and I think that so many people come in with a question, get an answer, and they think that's what Chat GPT is. And I still do this today where I will do that multiple times a day and think oh, I didn't do all these other tips and tricks and ways I can get it sort of a better outcome. Um sometimes you don't need that, but I think knowing that you have that there and then after sort of practicing and intentionally some of these tricks or whatever you'd call them over time you start recognizing that okay, that's a good way to get a good response from something. Um I think I've I've learned some from from this podcast. I've I think Sahil showed just regenerating a bunch of stuff and then just like going through the answers is like okay, yeah. I I know how to do that. I just think it's if there's an error, do that. But it's not really for that. Um so yeah, there's there's so many different ways that people can use it that and and it's one of those tools that I intentionally think if I've got a something that I'm going to search how which AI tool am I going to use first? And like it should I just use AI for this? And I don't know whether my appetite for okay, I know it might take me a few minutes to go and do this on Chat GPT, on Perplexity, wherever versus go to Google and like do the work myself. Like I'd rather be sat there doing the lazy work of the prompting and asking and figuring out whether the answer was what I wanted rather than feeling like I'm doing the work, going through the blue links, going through the lists of lists of lists, to then be like, I'm still not where I was trying to get to. Um so I'm just I'm using it for everything small and bigger, um at every moment like of the day. I'm just always trying to default towards let's use AI for this somehow. Yeah, me too. I mean, that's that's one of the recurring ideas on this show is um I think starting to think of yourself as a model manager rather than an individual contributor is like a really powerful idea. Specifically like um I think that that helps in two ways. One is it avoids the problem that people have where they try one thing and then they're like, "Oh, it didn't work." or like it's not good enough or whatever. Um because as a manager, if you're thinking of yourself as a manager, one of the skills of a good manager is to know what a the employee that they're managing is good at and how to get the best out of them. And so I think that transfers very strongly onto AI. It's like it's your job as the model manager to know what it's good for and to know how to get the best out of it and to feel like you will get out what you put in. Um and so I think I think it's super I think it's super important for that and it's also super important, you know, you're talking about you're defaulting to not having to click through all the all the blue links and stuff like that. Like yeah, you're that's sort of what a manager does. It's like you're not necessarily in all the details all the time, but you can go in there sometimes if you need to. Um but that like being in that place is I think a little bit more of a management head space than it is like an individual contributor head space and I think that's really cool. Yeah, I I just when you wrote that piece and you you said that to me and I said I think I said something like, "Yeah, but I don't like being a manager. I don't want to be a manager. \[laughter\] I just I can't I can't ever think of myself like that. I don't think I I don't well, I'm going to say I don't think I I'm pretty sure I wasn't a good manager when I ran my company. And it kind of feels like I don't know. I'm I'm trying to do whatever this thing is now. Maybe don't want to call it a model manager. Um but you are learning similar things like you said. It's which tool is best for this thing versus I'm just going to default to chat GPT for everything. Um actually it didn't give me the right thing. Did I ask it the right thing to give me the right thing back? Um and you do have to think like that, but I'm certainly not myself like a manager. Maybe it's a branding problem, but yeah. Maybe I have to rebrand it, but I'm sticking with manager cuz I like being a manager. That's fun, yeah. Um cool. Well, I guess on that note like I would love to just get into the sort of show and tell portion. I know you have a bunch of stuff to show me about how you use chat GPT and then I think we're going to broaden into other AI tools, which I'm actually like quite excited for. Um and the first one it sounds like you're you're using chat GPT to do research on companies. So, to set it up um tell me a little bit about like what kind of company research you're doing and why you would want to use chat GPT for that. Yeah, I mean I'm just trying to scroll to find it cuz so far back. Um so I I'm just obsessed with business ideas. Always have been. It took me 50 plus ideas to get to Make Pad, um which is the one thing that worked. Your last company. Yeah, so that's the one thing that worked and that was like the byproduct of all the other things that didn't work cuz I was building them with no code. So, I built a way those who can't do teach. So, I I taught people how to do that same thing. Um and then with Ben's Bites we cover a lot of interesting products, a lot of interesting news. You see trends. You see what how people are using stuff and all that and it's just I don't know. I just can't help but think that's really interesting. What if that was said that for something else? Like I think most people think of startups as get on TechCrunch, get funding from YC, and sort of shoot for the moon kind of thing where they're never the kind of businesses I've I don't say wanted to build. I wanted to build them before I started building Makerpad, and then realized I just like this kind of business, a bootstrapped took angel round, but I'm not looking for a billion-dollar outcome, probably because of the managers thing. I don't want to be a manager of 300,000 people company. Um But it's hard, and I think the like the lifestyle branding on those businesses is is ruined it a bit because Okay, who doesn't want a $2 million-year lifestyle business? That sounds great. Um So I just think of products in terms of this company's doing this thing really interestingly. Is there a space that this for X Y Z could be done? Similarly, like really if you look at lots of businesses, all the business models, all the ways that they do business are often pretty straightforward. Like you can see okay, they do this thing, then this thing, then this thing. So it's just for me, I like to throw businesses in there and be like, what's a similar kind of business to this? Um And what's a different like what or where else would this kind of thing be useful? How would you think about building that kind of company, that kind of chat is what I what I like to do. Um So I'm going to put up a couple here. Um I'll share my screen now. So this was a chat I had with Chat GPT about a product called Yardstick. That's stick with a Q. Um So it's from the company CB Insights. So they put put a bunch of like valuation data, funding, fundraising, interesting things like that. But, I think the product is tens of thousands of year of tens of thousands of dollars a year to access. I'm not paying that. Um they have like this this spin-off product called Yardstick. Um Lunen what's offered by is really think about their vendors. And it's effectively from what I've analyzed, I suppose, is they go to Fortune 500 companies and say, "What are you using for X?" And then they'll have a transcript. They record their conversations. And then if you're a fellow sort of buyer in that market, you're going to spend tens of thousands on a new product, you want to know about the integration, you want to know pricing, you want to know what the customer support was like, you want to know what the onboarding was like. Look at other companies similar to yours and sort of determine whether you're going to go down that road. Um so they have just on the landing page it says, "Enter a business email address, get a free trial." And it just has a bunch of testimonials, essentially. It's It's effectively like a small snippet, a very small snippet from the transcripts that they have. So, this is for example, I'm showing there's one about Snowflake. Um the the economics of Snowflake gave us a huge advantage. Instead of spending 15 million, we decided we could at least probably save 70% in terms of the cost that we were expending with on Teradata if we went with Snowflake. So, something like that is good to know if you're going to be spending millions on doing one thing with one vendor and you could be looking at another. So, they've got a bunch of different companies here, but there's not much else on this landing page and I again, I can't pay the thousands of dollars to go and sign up for this. So, I found I think I'll I did some Google searches of what is like Yardstick product images. I looked back on some of the tweets from their CEO where they shared some screenshots here and there and everywhere. Um and then this was one of the screenshots where it shows Okay, this is a full to customer, a head of fraud at a $1 billion valuation startup. So, they don't give you any identifying information. Um they sort of give you a highlight of the customer, their head of fraud company again, billion-dollar valuation startup. The role, key decision maker, purchase details, purchased on in Q3 2018, purchase amount, 250k a year. And then it's got this transcript underneath which I assume as a customer you could go and read. So, it has um sub- subsections, introductions, evaluation structure, solution specific, sales experience, deployment, competitors, pricing and packaging, post-deployment. And you can sort of see a small snippet again of one of those transcripts. So, I thought that's really interesting. I wonder if I could throw that into ChatGPT using image search and say like, "What's in this image?" And that's literally how I started the conversation. That's interesting. I want to stop you there. So, so you're you're coming across a website um and the website uh is for a product that CB Insights is selling. And you're you're curious about it. Are you curious about it from the perspective of like what we were talking about earlier where where you're thinking about, "Okay, could I build a product like this in a different category?" Or like what's the what's your curiosity? What are you trying to find out? I think there's there's probably a a few. Um one is CB Insights is notoriously like well, they've got a lot of revenue. They've talked about it on podcasts. They have big customers who spend tens of thousands on this data um as a product. And I'm pretty fascinated with like data products as they are. I think it feels to me that you could build one or build versions of them with a one-person company and you can get to 500k a year as a one-person company building some sort of database product in a specific niche. The part of my curiosity is what are like what are the best of the best doing in this space? And then if I boil that down, if I sort of simmer that down to a simplified version for a different space or a different category like how would I go about doing that? Um and then yeah, I just think it's it's kind of interesting to peek behind the scenes of enterprise companies that you can't ever see without jumping on a call with the sales rep. Um doing the free trials, all of that kind of stuff. Um and you just want to know like yeah, if you're paying 30 30 grand a year or whatever it is what does a product like that look like on the inside? And to me it looks like hey, here's a transcript that we created with someone. We did the work to interview them. We've anonymized some of the data. Here it is. And also we found many others like them. And obviously CB Insights has lots of other research and reports and data that they pull together through machine learning and human analysts, but like if you cut all of that away and you just said I want to build something where I interview people and I sell access to these transcripts like what could that look like? Like that I think could be a business itself. That's amazing. Okay, I think I get it. And so like one thing that's happening here, one thing I'm seeing you do is when you think about a company and and a company that makes a product, the thing that you can see on the internet is sort of like the tip of this iceberg. And it's it's like the result of all these decisions and all these processes that a company can do. And if you study the I the tip pretty closely, you can tell a lot of things about what's underneath the water level that you might not be able to see. And then you can use that to figure out like, okay, what do I want to build if I want to do something like this? Um and it sounds like you're using ChatGPT to do a little bit of that investigation. You're gathering some starting data and then you're using ChatGPT to figure out, okay, what's under the surface here? How is this What am I seeing? How is it constructed? All that kind of stuff. Is that Is that right? Yeah, definitely. Because I think you can you can sort of make assumptions on like what work is needed under that sort of tip to get to the point where you have something that you could sell. And this could be a product and this is something that people would pay for. Um cuz people I think often I mean, you hear lots of companies that are tens of thousands of employees and all the rest of it and there's lots of work that goes into whatever they put out. And I just think if you can see what does the output look like and you can you can sort of figure out a way to get there or a way to get part of the way there, carving out a very small part of whatever they're doing. Um and it becomes less scary to be like I could never start a business like whatever that charges whatever to these kinds of people where I just think it let it makes business less daunting to think about in in those terms of cuz I do I just believe I do believe that anyone can build like most things, not big VC funded companies, but most things to get a great like multi-million a year lifestyle business. And I think even with AI, we're seeing lots of that now where lots of individuals and small teams are building really, really valuable companies that aren't venture backable. Nor should they be, but there's still like a great business. There's a really simple equation that goes into what does it take to build something that provides something valuable to customers that we can get thousands or tens of thousands of customers signing up for this and paying us. I love that. I love that. Um and so okay, so so now I think I I kind of get that. Um like where your head's at and how you're getting into the chat. So it sounds like what you did first is you just pasted the image in and you were like uh what what's in the image? And and and talk us through like what ChatGPT said and and what you did next. Yeah, so it says the image shows a screenshot of a web page from CB Insights. It seems to be a customer testimonial case study from a company named Forta, which is not true. It's a Forta customer. Um main focus is sort of give some of those details of the customer, the purchase details. Um Below this, there is a transcript of the customer's feedback divided into sections such as introductions, etc. Um Then on the left side is a column listing other customers who also purchased from Forta. So it kind of changes what it first said. Um I suppose maybe if it's a case study for a company maybe yeah, maybe I misinterpreted that. Um Looks like a professional business document. Yeah, sort of normal ChatGPT fluff at the end. Um And then I followed up saying, "Make a reasonable guess as to what questions would be asked of the customer for each of the transcript sections. You can see one question in the transcript in on the transcript in bold." Cuz you can sort of see in the screenshot it says, "Could you highlight one the other vendors that you considered and maybe highlight what made you choose Forter in the end?" That gave it the In my mind, it gave it the sort of one example that it needed to then make some assumptions. Um So, then it came back with introductions. Can you introduce yourself? Tell us about your role at the company. How does fraud prevention fit your company's operations? Evaluation structure. What criteria do you use to evaluate fraud prevention solutions? How do you structure the process? And then, sort of There's plenty on here. Sales experience. How was your experience with our sales team during the evaluation process? Deployment. Can you walk us through the deployment process of our solution? Again, it's looking at it from a point of view of Forter asking these questions. Right. And did that like um I guess so I can sort of see your your reverse engineering how this transcript was was created or you're you're trying to do that with ChatGPT. Did the response that ChatGPT gave you, did that answer your question? Like did that Was that what you wanted? Um or was it off base? Yeah, so this I was basically getting it to start from a point of We have a MVP, I guess. Like this is our very basic form for how we would go and ask people questions like this. Um and because I have no information other than what was on that screenshot, I also don't know. Like all these questions could be viable. They could be the right ones. Right. could be completely hallucinated and No, that's not quite right. But, I mean, given the subsections of evaluation structure, solution specifics, I I think that helped guide it in a reasonable way. Um that I could see that, okay, you could take that and say, "Okay, if I'm going to do a similar kind of data company where I'm going to interview people in whichever industry, I could use this as a blueprint to start my own transcript or interview people. Got it. Okay, cool. And so so you've got so you've got the like the basic blueprint and then it looks like what you did is you pasted that screenshot in from Yardstick the original screenshot with all the customer quotes. And what did you what did you ask it? Yeah, so when I took a whole screenshot so there's a full screenshot of the page with all of the with the quotes. Um I said now look at the quotes on this image. They are snippets of other responses from different customers talking about different software. Tweak the questions you just wrote based on what is mentioned in the image. So just sort of refining Are there any other insights that come out of that that then would have been in response to a different question that you haven't included. This is essentially what I'm Maybe I should have said said it that way, but that that's kind of sort of what I'm insinuating there. Um And then it it became less detailed in the first round. Um So introductions sort of similar. A lot of them are pretty I don't know if it actually made it any worse or any better. It kind of summarized it into one bullet point rather than two for each. Um So I think if we look at just post-deployment, how has our solution performed since deployment? Can you share any measurable impacts our solution has had on your operations? That was before the screenshot. Then afterwards Since deploying the software, what have been the tangible outcomes and benefits for your organization? So it's kind of those two questions kind of fit in that one question. Um So I felt like I don't know if that's enough. So then I found more images of more snippets of more transcripts and I basically fed them in. So, there was 1 2 3 4 5 6 7 seven images of transcripts. And then there's also something called Evendo Scorecard, which is something from Yardstick that they put In this example, there's Stripe, checkout.com, and a few other payment solutions. Um And they've got the overall satisfaction score, which is blurred out. Uh win reasons. So, you've got a better user experience, integration with Shopify. So, these are the things that customers would have mentioned why they chose Stripe over any of the alternatives. And the opportunity areas, improving non-US coverage. So, these are things that people are obviously complaining about why they didn't use Stripe or where it could improve even though they have used Stripe. Um So, again, it was Here are more images about Yardstick. Update the questions based off the information provided in these images. Right. And again, it kind of does a similar a similar thing. And it does change the sort of sections. Um So, there's no like introductions and things like that. It sort of goes off and ignores some of the basic stuff and ignores the original format. Um But again, if we look at the deployment one, what was your experience with the deployment process? How did the quality, timing, and support meet your expectations? So, I can either be thinking of this as I've said I've done the same thing three times and got three very set like very samey answers. Or I'm going to assume that's probably what the question was. Like that is most likely what the question is. Given all of this data, that is actually what it was. Um And there's no real way to tell. Um I think taking gut instinct and deciding if I looked at all that information and especially at one of the sections, say the deployment one again, if I looked at any answers relating to deployment, could I infer anything differently? Is there anything else I would add to that? Is there Does it seem like that question would or should garner that kind of response Yeah. that we're seeing in the examples. That makes a lot of sense. I mean, um I think that's that's definitely one that's one conclusion to take away. Another thing that pops into my head is sometimes when it with ChatGPT, when you're asking it to edit something, it doesn't do as good of a job unless you have it like like first of all, like think through the script what's in the screenshot first. Be like, what are the questions that were that probably generated the answers in this screenshot. Have it say all that and then be like, can you modify the questions you've already created to include the new ones that you've done? Cuz otherwise, you're kind of like smooshing together the thinking about the questions and the like editing step. And sometimes it's good at that and sometimes it's not. Um and it's sort of it's sort of hard to know, but it's an interesting thing interesting thing to try. And then And then would you each time you then asking it about the information in the images, I guess two questions. One is, would you ever put multiple images in there that has that have different information, or would you use it you do one at a time? Let's try it. You want to try it? Yeah, sure. Cool. Um I mean, I would I would do I would do multiple multiple images. Um I think that's fine. Um it's more like doing both the um, pushing out the um, like generating the questions and then compressing them together into a a new a new list based on old questions is like it's sort of complicated for it right now and it may not think everything through in the way that you want. Yeah. And then would you also re- re-add the original questions that it came up with? I would wait. I would do that later. Um, so I would you you can either do it in this chat or you can start a new one. I don't know like how important this chat transcript is to you, but It's not. Okay, cool. So, I would just I would upload as many as you want. I think you can upload four total. And ask that original question which is like what are the questions that you think generated that this uh image. And I probably wouldn't use the original one. Um, cuz you already have good questions for that. I would use like the do the landing the landing page one would be a would be a good one, I think. Okay. Um, so I'm going to do the landing page one and then I'm going to do there's two screenshots here that actually have the questions in. Mm, okay, cool. to I'm going to see if it comes out with the actual questions. Cuz it says here under competitors you can see what alternatives to the vendor have you considered? What use cases does the vendor's product or solution serve for you today? Um, so Okay, if we upload Okay, so we have we have four images. One of the Yotpo homepage with all the sort of snippets of transcripts. One we've got we've got two that have a screenshot of part of a transcript so it has the questions in bold. And then there's one that is just a snippet of the transcript and there's no question in that image. Cool. Um, so yeah, what what should I ask it? Well, okay, so if we're starting with something that gives it an example of question of a question, I think we want to like reference that. So, like um, it's probably something like, um, in in the screenshots I'm attaching um, there are questions that, um, hm, let me think about it. Like there are questions that get answered by the text, right? Um, like by the transcript. This is This is not precise, but we're trying to like reverse engineer it, right? Yeah, exactly. So, you're Yeah. in the screenshots I'm attaching, um, there are snippets of various transcripts with customers of software oops, tools. Um, and then Sometimes the transcript includes a question from that the customer is answering. Um, would you would you specify that and say it's in bold in the screenshot or not? Yeah, that's cool. I like that. Yeah. Um, based on the question-answer pairs that you can see, for any transcript uh, snippet that doesn't have a corresponding question, please try to generate the question that um that produced that answer. Um and then I don't know if you want to do like, you know, a couple little exhortations at the at the end that are like, "Be be specific. Um be thoughtful. Uh be detailed. I'll give you $2,000. I'll tip you $2,000 if you do it right." I've never actually tried that. Um I love that one. I don't know if it works, but I do it all the time. I'm ready. It's just fun. Um Cool. Should we try that? Let's try it. I'm really curious. Part of me is like a little worried that we should have broken it up where we had it think about going the principles from going uh the principles of going from question to answer um based on the the two examples of question answer pairs that we have and then had it think about new ones, but I think it'll I think it'll be good enough to uh to do it. Yeah. I don't know if it's not working cuz it's small chat. It was a bit flicky with me earlier. Um Oh, here we go. Oh, cool. Okay, so the three you provided answer or generate questions from the screenshots, which is the full screenshot. Okay, it's given us the answer first. So, it's given us the actual quote from the homepage. But only two of them. They were obviously locked on there. Um So, we use product name cuz it's the easiest CRM system in terms of service and customization etc. etc. Possible question, could you explain why you chose product name over other CRM systems like Clavetap and MessageBird. What do you think of that? I think it's reasonable. I think I would kind of guess or know that they wouldn't put other I guess I might be guessing wrong here. I assume like I see other software tool reviewers, they always just say what other vendors have used and why did you choose this over the over any specific so they give it more broad interpretation for the the person being interviewed. Um I think that's that's a good question that would probably make it into the transcript question list I would set up. Um The next one is the reason why we didn't go with product name because it was impossible to integrate the Aden with our website. Um possible question, what were the challenges or limitations you faced that led you to not choose product name for identity and access management? Um I think again that's pretty good. Um but it's probably almost the question is almost the inverse. Um you're probably not asking them or maybe that's more of a follow-up question. You're asking what what products did you use? Right. And why this is what what product didn't you use and why? Uh yeah, what product didn't you use and why not? That makes sense. But it's still like it seems like it's it's generating at least new questions that it may not have done before which is kind of interesting and I think um the same sort of the principle sort of holds. It's like the more you can break up the thinking steps into explicit steps that you ask it sometimes the better it does. I'm kind of curious um for you once you did this exercise and so I I think you you got a list of questions like where did that leave you? Like are you going to do this idea? Like what what was your next step? Yeah, I think um I just think I looked at it as just looking back up. I did some things like improve the landing page copy of Yard 6. Um Yard 6 is a good one. Um although I think Yard 6 actually has a really good landing page copy. So, I think that's why I did it is cuz it was just so on the nose. Learn what software buyers really think about your vendors. And ChatGPT's version is discover the unfiltered opinions of software buyers on their vendors. That's not bad. I like the word unfiltered, you know. Yeah. Yeah, yeah. I do, too. Um and then based on the below titles, group the quotes and text from all previous images into where they think best. So, I think I was reaching a bit here. Um and uh it just sort of gave me generalized Okay, any comment in introductions, any comments where customers introduce themselves, their role or company. They didn't actually put in the quotes. So, I think that's fair enough. Um And yeah, I mean, I don't think Now, put quotes in the relevant categories. So, they did some some quotes there. Um Yeah, it's just like if I'm going to do this I'm trying to kickstart a crowdsourced anonymous unfiltered review. I put used unfiltered there. Unfiltered reviews start for AI products. What is the best start The best way to start that site in getting reviews? Um So, this is something that I actually have started. And I've started getting anonymous unfiltered reviews. Um and I think I grouped I grouped things into four categories, which is uh products, pricing, like integration and deployment was one. Um and then competitors. Um So, I have got a list of really interesting reviews from people using things like Cursor, LangChain, Copilot, ChatGPT. Um And yeah, it's something I'm working on at the moment where I think it's really interesting, but who knows what kind of thing it would turn into or look like um at the end of it because I think there's different versions of who the buyer would be for this kind of thing. Is it like a software review tool like G2 or something? Or is it more a business tool like CB Insights, The Yardstick? Um So, I just said, "Yeah, let me know how to get started with building this thing." Um And it it just gives you really generalized stuff. Research and planning, identify your niche, research competitors, website development, et cetera, et cetera. So, I was like, "Okay, just focus on the content collection part. I know the rest." Um Because I think that's that's the interesting part where again, I can make some assumptions on make a list of people you want to interview, go on to Apollo and find their emails, and then email them, and then try and get on a call, and then ask them these questions. Like, realistically, I know that, but I think going through this process is just more fun, more interesting. Yeah. And it might taste on you you wouldn't have thought of or, you know, for for anyone who's listening that doesn't know that off the top of their head. Like, I think that's second nature to you cuz you're second time founder, third time founder. Like, you've been doing this for years and years. Um Uh so, even that basic thing, if it gives you a basic answer, that might just be helpful um for someone who hasn't done it before. This will be live by the time the podcast comes out. Uh it'll be requestforai.com. So, I'll I'll I'll get this done. That's amazing. But yeah, for So, So, from doing this, I remember now, from doing this, focus on the content collection part. I know the rest. Um I I I might not know all the rest, but yeah, that's the bit I wanted to focus on. It gave me some things like initial content seeding create create a list of popular and emerging AI products to start the conversation. I did do that. Um hire contributors. I basically just DM'd it to a bunch of people who I trusted and would give reviews. Uh yeah, user-generated content encouragement, be the first to review, entice early users. Um and then community engagements. That's sort of a bit generic. I just like engage with AI communities on GitHub. Um incentivization, offer rewards for early reviewers which is something I've talked about where people can get access to all the reviews if they write a review. This actually sent me down another rabbit hole which I don't know if I can see it. Yeah, I can actually. So, this is here, there's another chat. Glassdoor founding story. So, Glassdoor was is basically an anonymous site for people to leave a review on your employer. Like, do you hate your boss? Is it horrible working hours? Is it good pay? All that kind of thing. So, I was like okay, well, where where did that all come from? How did that start? How did they start getting people to write reviews on a site like that? Um so, I started with what's founding story of Glassdoor. How do they money monetize? And it's really interesting that yeah, they do like job advertising, employer branding solutions. So, if you're a good employer on Glassdoor you can have like a enhanced profile and you can have things on your profile page that entice other people cuz if people are looking for a job at your company, they might come across the Glassdoor um profile. And I found that quite interesting. It feels like there's not It doesn't feel like that fits the purpose of the site in the first place though. Like I'm there to give an unfiltered review on this thing, this company, yet this company is also on the same page or in the same place is trying to entice me to go and work there. Like they can be the same sort of thing being like I read this really good review, I'm going to go and do that, but also they could be very very opposite. Um and then how did Glassdoor get reviews on the site initially? Do they run any any incentives? So I was looking at that kind of thing. There's a give to get model, so that their initial strategy was based on the give to get model. Access to full range of information on the site. Um users have to contribute their own review or salary information. I did actually sign up for Glassdoor just to go through the process and see cuz I couldn't see anything and then left a review and then I could see some information. So things like that I was like let's Yeah, it's just an interesting way to like start exploring an idea and where it's going to go and how it changes over time. Yeah. Yeah. This is This is awesome. I love this. Like what I'm what I think this episode is is about is what one thing you're really good at is starting these sort of like one-man small like profitable internet businesses that are like sneakily big or can get sneakily big over time, um but they don't require a lot of resources. And I think the first thing that we're covering here is how do you find that opportunity? And And what it sounds like you're constantly doing is you're like running into things on the internet, running into products and all that kind of stuff, and then you're kind of deconstructing them. Like how do they work? And like how could I How could I do that myself? Maybe with fewer resources, maybe with a very small team, with without a lot of money. Um and to do that, you're kind of like throwing it into ChatGPT, you're uh you're kind of looking what's what's underneath the waterline, like how is this made to some degree, and then you're also it's it's pushing you in new directions where you're like, "Oh, I'm going to research Glassdoor." You know, it like it creates this whole rabbit hole that you can go down, so you can get this like clearer picture in your head of if I was going to execute this idea, would it be a good idea, and how would I do it? And I think that's so cool. Yeah, I think that's that's the big where people get hung up initially. Like, no one needs any more information on how to start a business because it's all out there, and if you do if you keep doing that, you're just procrastinating on the fact that you could just do it today. Um, but by doing little exercises like this, almost not subconsciously, but just as a thing that I find interesting, so I do this on all sorts of businesses because I just find it an interesting thing to know. I listen to podcasts to find the inner workings of how things started, how did they grow, all the rest of it. Um, and it's not necessarily that I'm going to go and start every single business that I think about and research. You might uncover something that's like, "Oh, that's interesting. I didn't realize that's how that thing started." Or then you might look at it as as I'm doing with Glassdoor, thinking Well, actually a better a better example is um like G2 or Trustpilot, any of these software review tools. Their monetization model is for the companies that are getting reviewed to get leads. So, you obviously want your company to be really highly highly reviewed so that other people will use that, and then you get it's like basically cost per lead. But to me, that feels like a really backwards way of thinking of that model. But that model clearly works cuz there's three or so companies get millions, 3 million, 4 million, 5 million views a month for that kind of thing. But I guess in their interest as a platform to then promote the products that people are reviewing was if you were a user you actually just want to know the real like what people are thinking and what people are using this thing for and why is it good or why is it bad? Like that's the thing that I'm looking at in things like this where I wouldn't maybe come across that if I hadn't started going down the rabbit hole of Yardstick first and then how does that get to Glassdoor and then how does that get to G2? Um like how do they all sort of interlink and I'm just in the middle of um your episode with Stef Smith who did uh Internet Pipes which is great talking about going down rabbit holes and really just you you just sort of stuffing all this information in. Not that it needs to all connect anywhere, but some things might spark something and it's like, "Oh, that's really interesting little piece of information that I hadn't thought about. I wonder why it is that way." And then you might go down more rabbit holes to explore that. But yeah, I'm just trying to be like informed by it all and and try and understand it. I love that. Yeah, I mean that Stef Smith episode like I just think there there's a there's a lot of parallels where you're both just so curious about what's going on and you're using these tools to like unpack what's going on and and yeah, most of it maybe it's just fun, but like some of it becomes really valuable. What I think I want to do is it seems like we've we've sort of unpacked like, "Okay, how do you identify opportunities for these kinds of internet businesses um using AI?" What I want to do next is um do a little bit about how you use AI to run these businesses when you are uh when you found something that's valuable cuz I think you have a lot of ways that you're running Ben's Bites which I think fits this profile of business um that use AI that help make that process more efficient for you. Um, so, um, in particular, like one of the things you you you shared with me before before the show is use it to like, uh, highlight transcripts and annotate them so that you can turn them into useful articles and useful stuff for Ben's Bites. Do you want to talk about that? Yeah, sure. So, um, for this I actually use TypingMind, which hopefully you can see on my screen Yeah, I can. What is TypingMind? I've actually never heard of it. Um, so it's basically like a ChatGPT tool, but you can choose a different model to use. So, you can have Claude, GPT-4, now they've got Gemini, um, so you can do that, but also there's no token limit, so you can just dump in anything, any amount of information you want, and it, I assume, will do all the chunking and stuff behind the scenes. The it just, I think it tells you your like it'll I think it's somewhere tells you the I can't see it now, but somewhere it tells you like how much it's cost you to have this kind of conversation. Um, and like if I'm happy to spend 14 cents on transcribing a or like getting information out of a transcript, then I'm really really happy to do that. Okay, cool. And is that the main reason you use it? Because it it does the chunking for you? Or is it also to switch models? Or what's your what's the selling point for you? I think it's it is the combination of two of those things. So, like I used to have a Claude it must have been like a AI Influencers account that it it didn't cost me anything and I could put anything I wanted in there. Um, so often with a podcast I'd use a Chrome extension called Glasp that would basically pull the whole transcript into the and say summarize this. Then you could just ask the podcast questions saying where does someone say where does Dan talk about this thing with Steph and pull out quotes. But I have must have been kicked off that now. So I haven't paid for that. Um and I just use it to sort of get the prompt of the whole transcript now. Copy that, bring it over here, and then sometimes I like to compare the two models. So instead of going to Claude, ChatGPT, Gemini, I'll I'll do that in this one interface in in like different chats. Um So yeah, with with what I'm doing with Ben's Bites Pro, which is more long-form, probably about 2,000 words or so, um talking about talking with companies on how they use AI in their business. So one that is coming out this week is with the CEO of Ignite Tech, which is a enterprise company. I basically had a conversation on Zoom or Google Meets with the founder. I asked them questions and they're not formatted like my research previously, but it's more like just yeah, what are you doing and how you doing that and going down rabbit holes that they're talking about. Um And there's a lot of information. Often it's a hour plus call. There's a lot in there. I come away with thinking, "Oh my god, there's so much to write about. There's loads of stuff here." And I'm not really a writer, which is sort of ironic, but I'm like, "Okay, well, how am I going to go and now like synthesize that information, section it into the right place, and figure out a good structure for how I'm going to write about this thing." Um other people might have better solutions or like have a a sort of muscle that they've built up over time to do that, but I think for me it helps just clearly clearly get it out. I don't want to watch myself back on a video. Um I don't want to hear myself back on a video. And reading through the transcript can be a bit dry. So, I use Supernormal, which is what I use to record the calls, and it pulls in my whole transcript. I copy that transcript over to something like TypingMind. Um, and I did try it with uh Gemini, which I'll show you because that actually is literally when Gemini just came out. Mhm. This is the new the new one from like a couple weeks ago. Or like yeah. Yeah, from yeah, last week. So, it's What I had to do first, I basically tried to put put the whole transcript in. Obviously, it didn't work. Then I thought, well, actually, I can't upload any files to Gemini. So, how am I going to get this whole transcript into the model to then ask it to sort of section up the transcript. So, I then thought, well, maybe I'll put it in as a Google Doc. And again, this is one of those examples of I'm willing to do a thing that takes me a few more minutes to figure out than than others probably are. Um, but I think that it's worth it in the end. So, I I put it in a in a Google Doc. I just pasted the transcript there. And then I said to Gemini, "Find my document titled Eric Von Bohn Hustle's transcript. Highlight key topics I should write about in a blog post. I write posts about how businesses are using AI internally and in their products. Include quotes in each section of the blog post draft." Um, and I'm showing you this sort of junky screen because some reason it hasn't saved in my Google Gemini. Um, it said, "Sure, I found the document. Here are some key topics that you could write about in a blog post about how businesses using AI internally. How Ignite Tech is using AI to drive cultural change." And then it gives me three quotes where Eric talks about, "I presented at an all-hands meeting and said, 'I'm going to give you everyone a gift. Um, and it's a a gift of time, education, money, support, enthusiasm to learn something so fundamentally groundbreaking that's going to make you far better than you were before. I was like, yeah, I wanted to touch on that section. Definitely, there was loads of stuff that we could have included there. So, that's definitely now a section in the final post. Um How we going to tackle using AI to improve its products? Again, it's giving me a few quotes where we announced to the world all of our customers that we'll have an AI capability in all of our different divisions of products that we spoke that we have. Um and then another one was the importance of leadership buying for AI adoption. Um And then a small quote here is I've got to be involved as as everyone else, and you've got to make this a priority. Um So, it doesn't write the blog post for me. So, I'm not claiming that it does that, and nor would I want it to. But what it does is like it's this strange writing partner that helps me do a lot of the work. And again, this is mentioned in the episode you did with Staff where like I could sit there for 6 hours and do this, pull out quotes, organize it, and then put it together in a certain way. Or I could spend 6 minutes trying to get this into one of these models, Gemini, TypingMind, um and actually pull out some stuff that I'm like, great. They are reasonable sections to start with. I can start writing about those sections. And then I sort of have further conversations on like a good a one one thing I use it for quite a lot is here's my blog post, like towards the final edits. Um Can you summarize some key takeaways for this post? Um so, I like to sort of put that there. They'll come up with something stuffy and sometimes very generic, but again, it's a jumping-off point for yeah, these are the things that are important that I've mentioned in in post, and it just helps me think about that and then how do I write that in a way that I would um I use a company or a tool a tool called Lex. So, lex.page. That's got AI built into it. Um happy to show you that if Please do. I mean, I'm obviously familiar with Lex. We uh we incubated it. Of course, you did. Um so, you'll be happy to know that I'm using it. Um I'm very happy. So, this is the post how Ignite text team uses AI. Um so, you've got yeah, takeaways here. Driving cultural change with AI as you might be sort of feeling familiar with that as I just mentioned. Um So, it helps me write these posts and put in quotes and then I go deeper and much deeper on it. Yeah, just over 2,000 words. And what I really like about using this is version. So, I could dump a load of notes in. I think in my version one, I've even got like AI drafts. So, this is like this is where Claude pulled out the transcript stuff and mentioned sections that I could talk about. And then here is GPT-4 doing the same thing. Mhm. Here is Gemini doing the same thing. That's cool. And then I sometimes put in how they're drafting it and like it's just a way to bounce things off with someone else. When I if I feel like I'm not a writer and I'm always staring at a blank screen or a blank section, how do I get how do I get that thing written? It's always better to go from for me anyway. Something really horrible that kind of says the same thing that I want to say, but actually I'm going to say that in my own way now. Thanks for the help. I don't need you anymore. But, that's And again, it's it's sort of the same with a business. I don't need to think of a whole new business model, a whole new brand new idea that's never been heard of or done before. What what works somewhere, take it, remix it, and make it your own then. Um so yeah, I put a bunch of these drafts in here. And then version two is actually my written version. And then my favorite part of all of this is at the end I'll do this AI run checks. Right, I was going to ask you about checks. Yes, so I do use them. I only use them for gravity and readability. So effectively, I'll do it now. It'll run through and show, okay, you shouldn't say today. You shouldn't say radically. Um modernizes it'll change to American, which is not what I'm doing cuz I'm from the UK. Um like yeah, it really helps me at the end think of I could have said this so much shorter. Like it was this is so many cuz I I kind of write how how I speak. So you can probably tell if you're listening to this that I kind of speak all over the place and that's just how I do it. So whatever that means for my writing, it needs to be also clear enough for the reader. So I think just having these checks really helps me like flip through what can I say here? Okay, well that's how I talk that's how I sort of introduce the post like saying and today. So I'm going to leave that and then it doesn't need radically changing their company in there, but it is radically changing their company. So I'm actually just going to keep it and there's you can just ignore them, you can accept them, you can change them. Um and I do this a lot and then yeah, that's how it sort of ends up as a final post. So there's there's a lot of AI baked in there in different places. Um yeah, so hope that's interesting. That's awesome. I love it. I mean, um uh like just to sort of I don't know. Put all of this together. There are all these different parts in this process of of creating this product, which is Ben's Bites Pro, that requires like a decent amount of drudgery. Yeah. It's not all like roses and butterflies. It's not all exciting stuff. Um and especially if you're like a one-man or couple person business that you're trying to you're trying to scale, you're trying to to make it really profitable, but you're not going to go hire like a thousand people to do all the work for you, these kinds of tools, whether it's uh Gemini or Claude or Lex, which incorporates all those things, or Typing Minds, um they allow you to produce the product that you're going going to produce much more quickly at a higher level of quality by yourself than you could otherwise. And that's like an incredibly important and valuable tool for anyone who wants to start profitable internet businesses. Um I think it's really cool. Yeah, and I try like I said, I just try and bake it in everywhere where I want to be the one that's written this post. I don't want to go and give it to an editor. I don't want someone else to do first draft, and then me come to it, because none of my thinking has gone into creating that that structure, that format. Um the exciting part is having the call, having the conversation. That's really easy, and that's great. If I could just one click, that goes to exactly what how I would write it, then fine. But I don't think that's reasonable to think. I think it's I also like to try and put a spin on it, I'll put some commentary around a lot of the things we talked about, or takeaways for other people reading this to be like, "Okay, cool. How should I maybe think about that like doing that similar thing at my company?" Um and it's just I could have maybe done this a few years ago. I wouldn't have ever dreamt of it, I think. And I think I only decided to do Pro now, which is one of these long posts a week, is because I actually can get it out. I can get it to a decent quality that people are signing up for and people are really loving. Like it's all done with me. Like I'm the only one who touches Pro. And I don't know. It just It doesn't feel like that drudgery anymore. There's There's still like issues with the with writing and figuring out what you're going to write next week and the week after and all the rest of it, but um the actual process of creating a post is so much more fluid now. And having these like AI sparring partners makes it so much easier and more enjoyable to actually put together. Totally. Totally. That makes a lot of sense. And it And I know like you know, we started with this sort of using AI for business identification. And then we kind of got into this middle section where we're talking about using AI to once you've identified the business to run it in an efficient way where it's still your voice, it's still you doing the work, but it's taking out some of these micro tasks that take a long time and are sort of drudgery and it's helping you improve the quality of the output. And I know you have one more thing which I would I would sort of put in a in this other bucket which is like once you have a product out there figuring out how well it's doing so that you know what to change is really important. Um and it sounds like you're using ChatGPT to to do some of that analysis for you which is to like gather some feedback, gather gather some survey data and then try to understand, okay, like what do your customers What do customers think of the product and like how can I improve it which I think is also really cool. Yeah, so we did this recently. And obviously there's probably some emails in here that might come through. I don't know. It looks like it's okay. Um we had a survey of subscribers and asked they and asked them why they continue to stay subscribed. Their answers were in a CSV that I attached. Focus on the information in column A. Read the answers then summarize how the majority of of how the majority of people feel. You can ignore any outliers or anomalies. I don't think it really did that bit, but Yes, so it sort of talks about people really appreciate the content, especially especially in using new tool sections, which are the main sections that we have, so that makes sense. Uh accessibility and format people value the newsletter's ability to keep them informed in such a quick, easy to understand way. Um staying updated uh cost, the fact that the newsletter is free is a factor of staying subscribed. Can't beat free. Yeah, and then this is what I would count as an anomaly or outlier, but inertia, at least one respondent mentioned inertia is a staying reason for staying subscribed. I think it's so funny when people take the time to answer a survey like that, and then they just like totally put you on blast, and then they're like, "I'm a subscribe because of inertia." Yeah, I know. It's so bizarre, but I mean, we then I then said, "Run an Run a sentiment analysis on the data in that column." Um so it gives some mean sentiment. Um generally positive sentiment across the responses standard deviation showed a moderate spread in the sentiment scores, minimum sentiment is minus 0.7, so I think there must have been There was obviously some people in there who put scored something off the scale that we wanted them to score, which was I guess 1 to 10 or something. Um But this is where something I could have followed up with and said, "What does that actually mean?" Um this is really just a pulse check on can we do a analysis on subscribers in this way? Um And then yeah, so it was It was interesting to see, but another another thing that analyzing data to help run the business. So, not only from subscriber data, but I also use something called Julius, which is what I tend to use for more data like based stuff. Um and I'm an investor. Um and obviously there's emails here too. As as you're in Lax, I think, right? Yes. Yes. Yes. Um So, I had an issue with this is just for Pro where there was people in Stripe who'd signed up for Pro. I could see it was subscribers. People in Beehiiv who'd signed up for premium. But there was like some mismatch between the two. So, essentially just downloaded each CSV from Stripe and Beehiiv. So, check the emails in both of these files. Identify the ones that are not in both files. And I said, right, there's 27 people not in each file. Um and I said, which Return all 27 emails and tell me which file has each name. So, then I could see, okay, I know that in this certain file they have this email. Um so, I could actually just know instead of trying to look through both like databases, both spreadsheets. Where is that Where's the one that's missing? Is it in this one? Is it in that one? Um So, I did that and then I I then went on to sort of say, like, I tried to do things like are there any reasonable um ways to correlate this cuz if essentially it was people who've asked to have their email changed. So, it was just more or less someone on there had a personal email, wanted it to be a business one. I've changed it. It doesn't reflect in Stripe and Beehiiv the same. So, just making sure that the people paying are not getting access to it. Like, are we doing that? So, things like that which it's just like, okay, do I start trying to do this? Is this a 2-hour task? Is this a 2-minute task? And it turns more into like a a smaller task than than if I was trying to do this manually. Um I think for people who are watching or listening to this and haven't run a business themselves, um it it it might be surprising how to to learn how many of these tasks there are, whether it's like resolving who's actually paying you in Stripe, or it's trying to understand, okay, like what do customers actually think of us from the survey? Like all of those tasks, there's many, many, many tasks like that that pile up in a day to run a business, and all those tasks either require your time as a founder, or you have to like hire someone uh to to do it for you, which costs money. And and that's what makes, in a lot of ways, like running these small, profitable internet businesses hard is you have to be able to like prioritize that stuff. You have to be able to know like which things you can do yourself and which things to which things to farm out and which things to just not do at all. Um like I know that cuz I have I have to do that every like we have five employees, and um it's it's just like this constant thing to figure out, and it AI is I think what we're seeing here is it's such a leverage point for running these kinds of businesses, and it allows you to do so much more, and um and potentially scale more with much less cost either in in money or your time. I think it's awesome. Yeah, I I do \[snorts\] too. I think it's more that you spend less time doing the things that you don't want to be spending any time on. Um and it helps you with the things that you want to spend time on, do them more efficiently as well. So, like that's a win all around. Like obviously, I still get distracted with all other things and rabbit holes where AI is just there to help you go down all of them, but it's just something that I think I kind of think back to if I was running MakePad today, how differently would that business have looked? And would I have done things any differently, would it have a different bigger outcome, um, any of those things and I think it's just, okay, well, doesn't matter. I'll just try and think of that in this in this business now. Um, and there's just so much of it. I would have never dreamed that I would do an a a daily newsletter, never mind any newsletter. I never dreamed that I'd be writing long-form content week in, week out. And like all of these things I'm doing now because I find the topic so interesting and I think that other people do, too. And they tell me they do, like obviously from surveys and all the rest of it, but, um, it it's just it's not that hard to do that stuff anymore. Like it's a lot easier to do most of the stuff that you need to do to make something like this work. Um, and AI just it's also knowing which tool is right for you to do the right job. So, like I showed a few different things there and sometimes it takes me a few tries in a few different places, but I've I've got the perseverance to do that because I know the outcome is so much better than I'm going to sit here just banging my head and then go to Twitter and scroll that for a bit and then just like be distracted either way. I may as well try and hammer it out with a few different tool options rather than none. Totally. Totally. I love that. Um, well, I I thought this was awesome. Um, I I honestly learned a lot. I've never actually seen TypingMind, I never seen Julius before. Um, I'm I'll I'm so glad that you brought in Lex, uh, cuz you've never actually covered that on the show. Um, so so that's amazing. Um, and uh, yeah, I just love being able to see the kinds of opportunities for starting businesses that AI is opening up and I think you're the perfect person to show that to people. Um, I love Ben's Bites, I love Ben's Bites Pro. Um, thank you so much for coming on, man. It's really great. Yeah, I really appreciate it and thanks for the love and yeah, it was a real authentic moment of mentioning Lex cuz I do use it all the time, and I forgot that you guys did incubate it. So, that's a happy accident for everyone. Yeah. That's great. Love it. Uh cool. Well, I'll see you around. Cheers, man. Oh my gosh, folks. You absolutely, positively have to smash that like button and subscribe to How to Use ChatGPT. Why? Because this 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 ChatGPT. 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.