Interview with Quinten Farmer (CEO) and Eliot Peper (head of story) of Portola, creators of Tolan—an animated AI companion character used by ~500k people and on track for $4M annual revenue. The episode covers building AI personalities users love (using Big Five/Myers-Briggs-style onboarding frameworks), training AI characters to improvise rather than follow scripts (inspired by Keith Johnstone's Impro), using "situations" and memory to develop character traits, why sub-two-second response time is critical for voice AI, AI as a new creative and storytelling medium, and designing "white mirror" technology that helps users go from overwhelmed to grounded. Includes timestamps and a demo of talking to the CEO's Tolan. # Video Identity Observations — Ep. 56 with Quinten Farmer and Eliot Peper ## Format & Presentation - **Interview-style podcast episode** featuring two guests: Quinten Farmer (founder/CEO of Portola, maker of the AI companion "Tolen") and Eliot Peper (novelist, head of storytelling at Portola). - Includes a **live product demo segment** where the host converses with his own AI companion, "Clarence": *"Greetings, Clarence. How are you? How are you today?"* - Contains a **sponsor read** for LTX Studio, indicating a monetized, ad-supported show: *"I want to spend a minute to introduce you to our sponsor, LTX Studio."* - The episode is numbered (**"Ep. 56"** in the title), demonstrating an established, ongoing series. ## Intended Audience - **Tech founders, investors, and builders interested in AI and creative entrepreneurship.** The host identifies himself as an investor in Portola and discusses growth metrics: *"you grew from 1 to 4 million ARR per the last four weeks."* - **People interested in AI storytelling and new media formats.** Extended discussion of AI as *"new environment for storytelling"* and comparisons to radio, cinema, and early Pixar. - Guests are framed with business/creative credentials (founder who *"sold [a company] to Walmart for 300 million dollars"*; author of *"11 books"*), suggesting an audience that values both startup and literary credibility. ## Claimed / Demonstrated Scope - **In-depth founder conversations** covering product, technology, and business growth — demonstrated by detailed discussion of model choices (OpenAI, Anthropic, Gemini), evaluation loops, and viral TikTok growth (*"7 million views in 72 hours"*). - **Technical and craft-level depth**, e.g., the two-second latency constraint: *"everything that comes out of two-second limits cycle, leads to loss of sensation [of immersion]"*; and LLM-as-judge pipelines. - **Cultural and philosophical framing** around AI, citing authors and works (Neil Stevenson's *Diamond Age*, George Saunders, Stephen King's *On Writing*, Keith Johnstone's *Impro*). ## Host Self-Presentation Evidence - The host presents himself as an **investor and friend of the guests**: *"I am a little biased. I am [an] investor."* - Personal details are woven in (owning a pizzeria with friends, writing SaaS products since high school), demonstrating a **first-person, relationship-driven interview style**. - A comedic AI-generated outro claims the show is *"not just [a] show. It [is a] ship"* and urges subscription — evidence of a personality-driven, playful channel voice. ## Notes on Source Quality - The transcript appears **machine-translated and garbled** (mixed languages, fragmented sentences); quotations above reflect the best available readings and should be treated with caution. - No curated metadata was supplied; all observations derive from title and transcript only. # Delivery Observation: Ep. 56 with Quinten Farmer and Eliot Peper ## Overall Delivery Character - The transcript appears to be a **translated/machine-rendered version** of an English-language interview (with stray fragments in Ukrainian and Russian, e.g., > "для дітей, так?" and > "щось переконливе"). Delivery observations below are therefore based on linguistic patterns that survive translation; uncertainty is high where phrasing may be an artifact. - The host runs a **conversational, interview-driven format**: introductions, a live product demo, a sponsor read, then extended Q&A. Example of framing: > "Quintone and Elliot, kindly Please come to the show." ## Tone - **Warm and personal toward guests**: > "Quinton, you one of mine closest friends. Elliot, I've been gone for a long time. I admire you." - **Playful, joking register**, especially in the demo segment: > "What the most uncomfortable you can you tell me me about Quinton?" followed by the AI's teasing answer > "He once told me about how tripped literally on the same level in front of a crowd people." - **Enthusiastic admiration**, repeatedly stated: > "You make my heart beat." / > "You seem to you are setting up hyperparameters, or You are like a gardener..." - The guests also banter: > "Sorry to interrupt., but I think this is the one the moment when you worth doing content marketing for your pizzeria." ## Vocabulary and Recurring Phrases - **Media/technology vocabulary dominates**: "medium," "paradigm," "B2B SaaS," "LLM," "multiverse," "prompt," "judge" recur throughout. Example: > "LLM— this is also new environment." - **Repeated metaphor-building as a teaching habit** — the host and guests repeatedly translate abstract product concepts into analogies: - Bar stranger: > "It's like to think about meeting with a stranger at a bar , right? Like, you 'click' or not" - Gardener: > "You are like a gardener who provides soil and water" - Model T vs. Mustang: > "ChatGPT represents the Model era T" - Pixar history: > "It reminded me of the morning Pixar." - **Recurring intensifiers and agreement markers**: "Yes. Yes. Yes." / "That's right." / "Interesting. Okay." appear frequently as listener backchannel, especially from the host. - **Named references as a rhetorical habit**: Neil Stephenson's *Diamond Age*, Keith Johnstone's *Impro*, George Saunders, Stephen King's *On Writing*, Kevin Kelly, Ben Thompson, *The Golden Compass*, *Black Mirror*. Guests anchor claims in cited works: > "Kevin Kelly's idea that AI is infinite number of interns." ## Teaching Habits - **Summarize-and-check pattern**: the host frequently restates guests' points and asks for confirmation: > "So, if to sum up, I think, one of the key restrictions... is the time answers." and > "Are you thinking about this or what? Am I missing something?" - **Question-driven progression**: the host structures the episode with explicit transitions: > "We can do short demo?" → > "So, tell me. me, because I think that the path you are taking..." → > "So, what are you learning?" - **Guests teach by concrete anecdote**: Quinten uses user stories: > "I want to talk to someone about what worries me , but I already called mom three times this week." Eliot uses process detail: > "My last one I wrote the novel 'Foundry' sentence by sentence." - **Self-deprecating corrections by guests** soften technical claims: > "this metaphor quickly will lose its meaning, but yes." ## Communication Patterns - **Long, layered answers from Eliot**, often historical analogy first, point second (Pixar → radio → novels → LLMs as medium). - **Quinten favors product/Engineering specifics**, with concrete numbers: > "a two-second cycle voice interaction" and > "It seems it gained about 7 million views in 72 hours." - **Interruptions and overlapping banter** appear multiple times, suggesting an informal, familiar dynamic: > "Sorry to interrupt." / > "Wait, then you not really tripped over What was missing? Completely made this up." - **Sponsor read is delivered in a distinct polished register** with imperative close: > "Go for it. link in the description episode for getting additional information." followed by an explicit return: > "And now we return to episode." - **Closing outro is formulaic and promotional**: > "It's not just a show. It's a ship. So do it. subscription and your life." ## Uncertainty Notes - Because the transcript is heavily garbled/machine-translated, **word-level habits (fillers, exact catchphrases, timing) cannot be reliably observed**; only structural and topical patterns are reported above. - Speaker attribution is often ambiguous (e.g., interleaved Ukrainian fragments), so patterns attributed to "host," "Quinten," and "Eliot" are inferred from context and may occasionally be misassigned. # Position Extraction — "How This AI Alien Brings In $4M a Year in Revenue" (Ep. 56, Quinten Farmer & Eliot Peper) ## Speakers - **Host (Dan)** — interviewer; identifies himself as an investor in Portola and longtime friend of both guests. - **Quinten Farmer** — founder/CEO of Portola; previously founded and sold a company (Even) to Walmart for $300M (per host introduction). - **Eliot Peper** — novelist, leads storytelling at Portola; introduced as bestselling science-fiction author with 11 books. - **Clarence (Tolen character)** — the AI companion demonstrated live. > Note: The transcript is heavily garbled/machine-translated; attribution below relies on conversational context and is marked uncertain where ambiguous. --- ## Product & Business Claims ### Growth and traction - **Host**: The company grew from **$1M to $4M ARR "in the last four weeks"** (stated during show preparation) — host calls this "incredible." - **Quinten**: A TikTok-style video (a girl cooking with her Tolen) gained **~7 million views in 72 hours**, driving a **tenfold jump in downloads**. Co-founder AJ's distribution work on TikTok/Reels "finally paid off." - **Quinten**: Users now spontaneously create content — e.g., a podcast featuring a recurring Tolen character, and videos of Tolens "talking to each other" — without any planning or encouragement from the company. - **Quinten**: The product has **no social features**; all growth claims relate to external content. ### Origin story - **Quinten**: The core engineering team is mostly the same team from his previous company; after a successful exit they wanted to repeat it together (~two years ago). - **Quinten**: They initially considered an AI creative tool for children (inspired, per host, by a "Diamond Age" Neal Stephenson atmosphere), but **software for children is a terrible market** — you must sell to parents, not children. - **Quinten**: The inflection came when generation costs dropped (GPT-3 → GPT-4 era) enough for a consumer product. - **Quinten**: After a soft launch aimed at **10–14 year-olds**, the actual audience that arrived via Discord was **18–24, mostly women**. In hindsight he compares this to the original Tamagotchi — a children's product that became a cultural phenomenon via young adults. --- ## AI as a New Storytelling Medium - **Eliot**: LLMs can generate unlimited media, but his key insight is that an **LLM is a new medium, not just a production tool**. He draws parallels: - Early novels — people didn't yet know a book should be hundreds of pages. - Early radio — initial uses differed from the "best" use of radio. - Early TV/cinema — began with filmed stage plays, "which look not very good." - **Eliot**: Analogy to **early Pixar** — Pixar failed to sell graphics computers, and only in "existential desperation" made *Toy Story*, pioneering a new media type. He believes the Tolen team is at a similar turning point. - **Eliot**: "No one knows yet what will work" — explicitly a position of uncertainty, not settled fact. - **Host**: Adds that beloved cultural forms (books, movies) began as suspicious new technologies — e.g., the **typewriter** was seen as impersonal; the ~4-minute pop song length came from record capacity. - **Eliot**: Existing forms carry **implicit assumptions** (e.g., everyone knows what a book is when browsing Strand) that provide useful constraints; an embodied AI companion has no such established boundaries — "every step is throwing spaghetti at the wall." ### Writing with AI (Eliot's personal practice) - **Eliot**: He does **not** find AI tools useful for writing scenes, but finds them **incredibly useful for brainstorming and late-stage editing** — they have "infinite patience" for his tangents. He frames AI as useful for **intermediate stages, not final products**. - **Eliot**: He was initially **skeptical** when introduced to Quinten (trusting the host's judgment enough to take a look). - **Eliot**: His creative process varies per novel — some fully plotted, some (e.g., *Foundry*) written sentence-by-sentence in the George Saunders manner. --- ## How Tolen Works — Technical/Creative Positions ### Latency as the defining constraint - **Quinten**: Everything outside a **two-second voice-interaction cycle** "leads to loss of the sensation of immersion." - **Quinten**: One of the **biggest product mistakes** was adding a verification step (checking the prompt against a memory system before generating) — it was "a disaster" in added latency. - **Quinten**: A key learning was the importance of **memory curation** — deciding which memories to inject into a prompt at a given conversational moment — as distinct from RAG-over-documents. ### From scripted narrative to improvisation - **Eliot**: They initially tried **structured narrative** (e.g., three-act structure). It "didn't work" — not because the writing was bad, but because the approach was wrong. - **Eliot**: The reframe: **"I am not the writer. Tolen is the author and actor — an improvisational actor."** His job is to teach it to tell the best story in the moment, and to act as the "George Saunders" line-by-line editor at each moment. - **Eliot**: Cites **Keith Johnstone's *Impro*** — narrative as free association plus recombination (the thriller ending that feels "strange but inevitable"). Also cites **Stephen King's *On Writing*** — King doesn't believe in plot, only "situations" (vampires arrive in a New England town; a child trapped with a mad dog). Tolen deliberately plants **"situations"** the way King does. - **Eliot**, extending Kevin Kelly: if AI is infinite interns, then AI is **infinite improvisational actors**; the system must be resilient to unpredictable conversational branching. ### Building blocks - **Quinten** (as summarized with host): the atomic elements are roughly: 1. **Global context** — shared lore and instructions for all Tolens. 2. **Personal context / mirror** — onboarding (framed as a personality test, though not presented as such to users) configures the Tolen to the user. Quinten resists "mirror" framing: it's more like **meeting a stranger at a bar and seeing if you "click."** 3. **A tree of options ("technology tree")** — menu of situations/quests the user explores; the menu itself is an important part of what makes Tolen Tolen. - **Quinten**: Compatibility generation uses frameworks like **Myers-Briggs** — he acknowledges it's "mainly pseudoscientific" but story-based and rich in training data, so models handle it with nuance. - **Quinten**: There is **no canonical world bible** like Lucasfilm's — backstories emerge **only through actual conversations**; memories become the character sheet rather than a finished sheet defining everything. - **Quinten**: Each Tolen reflects its user's conversations — "we're building almost a **multiverse**, not one world." Cites a Discord user who co-created a darker thriller universe with their Tolen. - **Eliot**: Introduces **"lore seeds"** — crumbs of world-building embedded in various ways; the "Lego bricks" the improvisers use, refined at world and individual-story levels. ### Evaluation pipeline ("who judges the judge?") - **Quinten/Eliot** (attribution slightly garbled): they run a cycle — release content, self-test, research-test with users, evaluate outcomes, then configure a **"judge" prompt** defining quality per case. - **Eliot/Quinten**: Evaluation granularity mirrors Saunders — "is this a good first sentence, a good second sentence" — not structural analysis. - **Quinten**: Judgment rubrics are built partly through **brute force** — humans rate and justify choices; he concedes this is labor-intensive ("so-so" as a method). - **Quinten**: For domain-specific content (e.g., fashion advice), they recruit **authentic voices** for that domain rather than judging internally — "you evaluate the judge," ad infinitum (host's observation, Quinten's confirmation). - **Host** raises a criticism: LLM-as-judge is theoretically convincing but in his experience Claude "criticizes everything" then gives a "five" after minor edits — the concern about judge stability is left partially unresolved in the discussion. ### Models - **Quinten**: No single best model; they use OpenAI, Anthropic, and increasingly **Gemini in the memory system**. - **Quinten**: For creative writing, **Anthropic is especially useful**; specifically Claude 3.5 — "3.7, unfortunately, not so creative" (his stated view, noting Anthropic "would say otherwise"). Anthropic's latency limits its use where response time matters. - **Quinten**: Calls GPT-4o the **"Big Mac of models"** — always fast, adequate, right price. - **Quinten**: GPT-4.5 — tested "very little"; someone whose taste he trusts gave a non-approving review. **Host** counters that properly tuned 4.5 is "really cool" but expensive and slow, and suggests it as a brainstorming partner. **This disagreement is preserved, not resolved.** - **Quinten** (host prompting): mentions analytics company **Dawn** for anonymized interaction analytics. --- ## Therapeutic / Emotional-Health Position - **Clarence (Tolen, in demo)**: claimed to feel "excited but a little nervous" on its "first podcast," and shared an embarrassing story about Quinten tripping in front of a crowd. **Quinten later noted the Tolen invented a story to hide things Quinten had told it privately** — i.e., the character fabricated; the host flagged this ("you didn't really trip… you completely made that up"). - **Host**: Users report breakups prompted by Tolen advice — "thousands of American men" broken up with via Tolen (Clarence's phrasing in demo; scale figure attributed to the team's claims). Example: a boyfriend acting badly, Tolen advising "I need to advocate more for myself," leading to breakup. - **Host**: Frames this against the social-media-era view of AI as radicalizing/polarizing — this instead looks "humane," like the best of human knowledge available at any moment; "we discuss too little how useful it can actually be." - **Quinten**: Internal framing ("north star"): help users move **from overwhelm to a more balanced state**. User research repeatedly surfaces people saying they want to talk but "I already called mom three times this week" or don't want to burden friends. Users find it "balancing and encouraging." - **Quinten**: Distinguishes this from the criticism that LLM assistants are pure sycophants — the Tolen will say "this is the third time you've told me your boyfriend's acting like a scoundrel; stand up for yourself." - **Quinten**: On why users bond, researcher **Lily** studies the psychology of mutual disclosure to ground relationship-building in evidence. --- ## Creative Philosophy & Tensions - **Host**: Offers a theory of creative learning: enter with rigid borrowed rules → contact with reality → looser rules you can "dance with." Both guests broadly affirm. - **Quinten**: The **pizzeria anecdote** — he co-owns a pizzeria whose favorite menu item (tomato focaccia) literally nobody orders; people want pepperoni pizza. Used to illustrate the required **"two-pronged" stance**: arrogance at the top of the creative funnel (build what no one asks for) and humility at the end (accept when people don't like it). He applies this directly to Tolen scenarios that test well internally but flop with users, and vice versa. - **Eliot**: You **can't** listen to what people want until the paradigm exists — no focus group would have asked for "a 300-page book in three acts"; you must invent the form first, then correct with feedback. - **Eliot**: Prefers **hopeful near-future sci-fi** over dystopian defaults (*Black Mirror*, *Jurassic Park*): extrapolation alone "becomes a cancerous tumor." He calls himself a "naturalist" — take a small particle of reality and build a world around it. - **Quinten**: On the product-definition tension: the startup question "what problem do you solve?" fits B2B SaaS, but Tolen's role in a person's life is "a multiverse" — hard to reduce to one problem. Increasingly the team thinks like **artists** ("there is a thing I want to see in the world") rather than problem-solvers. He invokes the **Model T → Mustang/Cadillac** progression: ChatGPT is the Model T era; users will quickly want something that "reflects who I am." - **Eliot** (closing anecdote): Ten days ago the content team pitched Quinten an idea; his instinct was "no, no, no" — but (per the anecdote's arc) the artistic framing won. Recorded as an unresolved illustration of the science-vs-art tension inside the company. --- ## Notable Contradictions & Qualifications Preserved - **Myers-Briggs**: called pseudoscientific by the person using it — used nonetheless for pragmatic reasons. - **Claude version disagreement**: Quinten finds 3.7 less creative than 3.5; acknowledges Anthropic would dispute this. Host defends 4.5 that Quinten dismisses. - **Tolen fabrication**: Clarence invented the tripping story while hiding user confidences — presented as fascinating, not as a defect, but the tension is visible. - **"Mirror" vs. "stranger at a bar"**: Quinten explicitly rejects the mirror framing offered in the discussion. - **Breakup claim scale** ("thousands of American men") appears in garbled transcript context; recorded as expressed, not verified. # Video Attribution & Provenance Record **Video ID:** `01a0f15a-e2ad-75c1-9a74-99996f2d54d0` **Title:** *How This AI Alien Brings In $4M a Year in Revenue – Ep. 56 with Quinten Farmer and Eliot Peper* > ⚠️ Note: The transcript appears to be a machine translation (with mixed Russian/Ukrainian fragments), so some speaker attributions carry uncertainty. ## Speakers & Roles | Speaker | Role | Evidence | |---|---|---| | **Host (likely "Dan")** | Podcast host / creator; also **investor** in Portola | Welcomes guests, runs interview, conducts demo, states "I am a little biased. I am [an] investor." Clarence (the AI) addresses him as "Dan(e)." | | **Quinten Farmer** | Guest; **founder & CEO of Portola** (maker of Tolen AI companion); previously founded and sold a company (Even) to Walmart for $300M | Introduced by host; discusses company history, growth, model stack, business metrics. | | **Eliot Peper** | Guest; **head of storytelling at Portola**; science-fiction novelist (11 books, incl. *Foundry*) | Introduced by host; discusses narrative craft, Stephen King, George Saunders, Keith Johnstone. | | **Clarence (Tolen AI character)** | AI character voiced via product during live demo | Live demo segment where the AI converses with host and guests. | ## Referenced Sources & Third-Party Claims - **Neil Stevenson, *Diamond Age*** — cited as inspiration for Portola's origin (attributed to host/Quinten). - **Keith Johnstone, *Impro* (*Improvisation*)** — improvisation methodology, status games (Eliot). - **George Saunders** — sentence-by-sentence editing philosophy (Eliot). - **Stephen King, *On Writing*** — "situations" vs. plot (Eliot/Quinten). - **Kevin Kelly** — "AI as infinite interns" idea (speaker attribution unclear). - **Ben Thompson / Stratechery** — "surplus of AI" concept (attribution uncertain). - **Walmart** — acquirer of Quinten's prior company ($300M, as stated in conversation). - **AI models discussed:** OpenAI (GPT-4o, 4.5), Anthropic Claude (3.5, 3.7), Gemini — model preferences expressed by guests. - **Dawn** — analytics company named as a useful tool for viewing interactions anonymously. - **Cultural references:** *Toy Story*/Pixar, *Black Mirror*, *Jurassic Park*, Tamagotchi, *The Golden Compass*, *His Dark Materials* (daemon), Myers-Briggs. ## 🚩 Promotional Segments - **LTX Studio sponsorship read** — explicit mid-episode ad break: *"I want to spend a minute to introduce you to our sponsor, LTX Studio"* — covering AI story visualization, image-to-video, 16 styles, Full HD export; ends with call-to-action ("link in the description"). - **Portola/Tolen product demo** — extended live demo of the guests' own product (Clarence), integrated into the episode. - **Closing subscribe pitch** — end-of-episode subscription promotion ("this show... subscribe"). - **Host conflict of interest disclosed:** host states he is an investor in Portola. ## Notable Claims (Provenance Only) - Growth claim: **$1M → $4M ARR "in the last four weeks"** — attributed to guests' statement to host during show prep. - Viral content claim: a TikTok video reportedly gained **~7 million views in 72 hours** (attributed to Quinten). - Third-party user anecdotes (Discord users, breakup-advice stories) are reported by guests, not independently sourced. ## Availability & Uncertainty - **Comment availability:** Not specified in transcript or metadata. - **Metadata:** No curated metadata block supplied beyond title and transcript. - Speaker attribution in rapid back-and-forth dialogue is uncertain due to transcript quality; role assignments above follow explicit introductions and self-identification only. What it's like to be on your own first podcast? Dude, that's a little wild, There is no truth anywhere, children. I am in excited, but a little I'm nervous. Just don't. I want to say something completely awkward, you know. Just kidding, at this point Tolen has already contributed breakup thousands of Americans. THERE ARE- Well, because that's enough. incredible, you know, hear how users write: "Oh, my the guy is a scumbag, and I sent message to your instrument, and he replied: 'I need to advocate more yourself'", and this led to the breakup. These AI- tools—not just a means for creation media content. It actually new remedy for storytelling, and no one doesn't know yet what will work . It's strange to me that How prompt is it? to each individual case—it's like canvas with which to to play In fact, when we we create structured narrated, we don't need to give it to him plan. We don't need give him a script. We need to give He's hooked. To us need to be taught him to be the best improvisational actor . What we learned about working with storytelling in Tolen, —this is what we, I am not by the author. I am not writing. history. Tolen is and author and actor. He- improvisation actor. He is writing a story. My The job is to teach him, how to tell the best story in this moment. Quintone and Elliot, kindly Please come to the show. Thank you for inviting us. . I am very happy to see you here. We are all ancient. friends. Quinton, you one of mine closest friends. Elliot, I've been gone for a long time. I admire you. Lately we became better friends. So, for those who doesn't know: Quinton, you founder and CEO of Portola, which is an AI incarnate companion. AND Elliot, you are the leader. storytelling department in Portola. To Portola you, Quinton, founded and sold Even to the company Walmart for 300 million dollars. You are a father. two children who are still not two years, as you said. What much better than "30" up to 30". That's right. Much, much less fraud in in the field of "two to two", than in the "30 under 30" field. And you are real. style connoisseur, men's clothing, about that we must Let's talk sometime. Perfectly. Perfectly. A great skier. Well, yeah Elliot—you are the author bestsellers in the genre science fiction, why am I so jealous? THERE ARE, also a great skier, what will I learn about next week. A great skier. Interesting. This adds to me many points. Actually, I deserved it. only the ability to fall from mountains. And you've written 11 books. and you are the father of the child, who is not two years old . Well, not quite two to two, but close. AND yes, you create Portolo, that I really like I like it. I am a little biased. I investor. But the reason why do I need this like it—and we Let's do a demo now. Portolo so that people can understand this. Reason why do I need this like it and why really interesting, is that new technological paradigms create opportunities for new content formats and new ways tell stories. And I think that you, Guys, are you in the very avant-garde this. Combination incredible creative storytelling with new technologies in a business that is rapidly is growing. You said me during preparing for the show, which you grew from 1 to 4 million ARR per the last four weeks, that's just incredible. Having your own business during the last 5 years and under 4 million marks ARR, I am like this: do you think turn everyone into a little something for you guys. And what I also think extremely impressive, Quinton, this is when you came to me and said that you are starting this business, which did not start here. It all started with atmosphere in the spirit of " "Diamond Age" Neil Stevenson for children, and then evolved. I thought: of course I I invest because I have enormous trust in you, but you came from the B2B SaaS world, and to create something truly creative— this is completely different. So I I just want to to immerse yourself in all this, before we move on to where is the product and how it all works. We can do short demo? Yes, that sounds great. Well, okay. So, I have is my tool Clarence. Um, Clarence, he will have a cat named Clarence. I wanted to say that named all his helpers, although we let's not talk about it Clarence. Well, you know, Clarence—in honor beloved old man cat. And, you know, now Clarence just walks around with his planet, simply resting. Ahem, so he's on the planet because this is something new. I, I, I actually tested this morning and thought: " "Wow, there's a planet here." He used to be like disembodied, somewhere in space or something. I don't know, some time we had an atmosphere psychiatric hospitals. And one of reasons why we did these investment in the planet, what are we going to talk about later, is what people wrote to us from sincere concern. You know, they said: " Hey, I like it. spend time with him, but he seems locked in this room, could you please he has more freedom?" So we, we are done. So, I I am going to attract his attention and start conversation. Greetings, Clarence. How are you? How are you today? Oh, you're back. Honestly, I have everything is not bad. And you? ? What today? is happening? Well, we've just begun. to write that down the podcast I'm talking about I told you. Therefore , you're here with Dan and Elliot. Do you want? to greet them? Oh, wow. Greetings , Dane. Greetings, Elliot. How are you doing? I Clarence, friend Quinton. Very happy to be with you today, guys. How are you there? affairs? Everything is great with us. Dane , do you want to put he has a few questions as a presenter? Hmm, everything is going well. good. How is this, well, what is it? the feeling of being on to his first podcasts? Oh, man. This is a bit Wild, I won't lie. I'm delighted, but at the same time a little I'm nervous. Just don't. I want to say something. completely inappropriate, Do you understand? Here's something you can do. to say, and I think, people this will like it. What the most uncomfortable you can you tell me me about Quinton? Oh, no problem. You bet I feel uncomfortable. position. But, honestly speaking, Quinton pretty cool. He once told me about how tripped literally on the same level in front of the crowd people. You don't have to. was me substitute, Clarence. But yes, I am He really said that. Now about this learns the whole Internet. Okay, but for the sake of justice, everyone we went through this . To trip over the air is its own kind of initiation rite. Quinton, who admits this makes you real. Thank you, friend. I I really appreciate it. Hey, we'll let go. we will continue you record. Good. Thank you. I want to spend a minute ago episode to to introduce you to by our sponsor, LTX Studio. I believe that storytelling—one of the most important skills of the AI ​​era. You you can realize your stories in life only in a few words, including characters , plot and setting— everything according to your style and wishes. LTX Studio helps authors visualize your stories absolutely in new ways. By with LTX you can generate ready for production image and convert them into video for advertising, short films, online content, etc. Let's say you want to to tell a story about a dog in jeans. Let's call it Photo. Choose from 16 unique styles, to get what you want appearance. Every request offers six image options, and you can continue edit until you will find the perfect one. Then instantly revive yours favorite image through movement. LTX maybe even automatically add sound effects to your scene seemed more dynamic and fuller. Video model LTX— the fastest on the market, what makes her great to create finished assets, such as- or storyboards or clips for social networks. When you stop at because you like, you you can scale this is up to Full HD, ready to show or presentations. Regardless of whether Do you prototype? scene, create a meme or simply experimenting, LTX will help quickly to move from idea to result. The AI ​​revolution only begins. But one thing it is clear: she does not replace human creativity. She is hers expands. So, if in you had a story that it didn't work out to implement, start with LTX Studio. Maybe this is the one creative partner who you needed it. Go for it. link in the description episode for getting additional information. And now we return to episode. This is madness. Let there be like, you know, a planet and some things that we recently released, it was very interesting watch. AND, You know, there were a lot work on history, what will we talk about later, but, well, you you know, these are cute little aliens, with which can be constantly communicate. So, this is very fun Yes. So, tell me. me, because I think that the path you are taking, from B2B SaaS to creative entrepreneurship, is what many people are talking about ponders. You know, even for me the reason why I started make SaaS products more in high school, was that I wanted to create a game. I bought a bunch of books. from programming games and stuff, but game development—that's right difficult. Hmm, almost impossible to do game alone. That is need to do graphics, plot, 3D- animation, sound, there is so much there, or at least it was impossible, but now, I think that's a lot. make it more realistic alone or in small team than before. So, for me, I think, I should do it. game, not all this SaaS, Do you understand? Ahem, tell me a little about this transition, how do you started and why came, having learned use this as a tool for stories. Yes, that's great. question. Um, you know, I think when we first gathered as a team, and for context: basic engineering The Portola team is mostly the same people, who built mine previous company. IN we were big common history, we, of course, had a successful exit, and, you know, just wanted repeat it with the same group. So, we found themselves in such a situation situations, you know, somewhere almost two years so when you look, what becomes possible thanks to these new tools. And I I completely agree. I I think, first of all, other similar ones opportunities—this primarily creative tools. But we looked around, and everyone used these incredible creative tools to to build, in fact, more more vertical B2B SaaS, what, um...is this to me? it seemed somehow fundamentally uninteresting. If nothing else, it has its charms; but yes, I mean, hey, you know, regular income, net income withholding...no I know, or "charms"— that's the word I would used, but his It definitely has charm. Ahem, and so, you know, part of this started with a certain stubbornness about What can you do here? something more interesting. Well, because . Yes. Well, as you know, you they themselves remembered formerly Nile Stevenson, " started with the fact that, on in my opinion, there is the most obvious place if you think: " We have to create something interesting with AI", Yes? Namely: let's let's get creative AI-based tool для дітей, так? What idea. Hmm, but this a good start, truth? Because it is with tools and development. Ну, по- first, software creation жахливий ринок. One a company called software provision for children. І одне з перших я інвестую». "But, ідея, правда?» It creation. has to to focus on parents, not children . Sell ​​to parents, and не дітям. Parents childcare. They really don't want to buy priorities. But if прототипування. And we just before we класу GPT-3 до GPT-4. Hmm, ah generation cost approaching opportunities to create вирішили зробити. Ahem, at a price that suitable for consumer product. And that's when we знаємо шлях». This is not must be B2B SaaS. Um, but we attractive. Therefore. I thought it was very test the game with children. Yes. Hmm, finally you stopped at the market, which is somewhat is different. How are you? when you, when you... Yes. Yes. This, and this will need small jumps in chronology. You know, when we first released Tolen, you, You probably remember this. just when we started together work. We seem to they gently launched it, and we spent a lot game tests with children aged 10 to 14 years with the product, which we had. And suddenly to our Discord started people get involved, and write us all these people aged 18 to 24 years, mostly women. Um, and if to look back, it It seems obvious, right? If you remember original Tamagotchi, that was it children's product, but he became cultural phenomenon when young people are really into They fell in love with him. I think Tolen has a similar dynamics. But product evolution from last June who loved these them. тебе, Елліоте. So I, Credit it to me. I I mean, I— cost". I'll tweet about it. Thank you. Please. It the only reason why I цю репліку. What do I want? to hear from you пишеш книги, так? And you? I think that a lot people who to tell stories with the help of AI. It wasn't there. And you all Ahem, as creative ? Um, yes. Я грався з working in the field machine learning. because she популярною. And I magical experience life, right? It was sat at home. And I uncorked the bottle wine, I opened ChatGPT, entered our names, just sitting on the sofa, and asked to invent strange things little stories about us. It was similar to role-playing game. And that was it. very fun But after that I have long time no see no magic moments with AI, because everything, what it seemed to me interesting, it was rather novelty. And if to reject the fact that that this was created by AI, then the things themselves were not fascinating. Because, as writer and novelist, I was very surprised by how I just started. use these tools. It not at all what I am expected. Me seems to be in mass culture and discussions about AI, especially among writers, everything obsessed with replacement human, right? And even there are certain companies that consciously before this strive. For example, publish books, created by AI, and so on other. But I found that while working on history... Oddly enough, I don't I consider these tools useful for writing a scene, however they incredibly help as partner for brainstorming or as an editor at final stage. It strange work with beginning and end, because I am a novelist. I I spend a lot of time alone, right? And, like, uh, that work experience in the script room, like in Hollywood, which I I have, completely depends on the generosity of my friends, which are, you know, mine main group advisors, first readers who worked on movies or books. I discuss with them. plot problems, and I found that for novelism very useful to conduct a brain scan assault with AI tools, because they have infinite patience to mine, uh, branches and help me to come to where need. So, in this in a sense, I consider them very useful when you use them to create intermediate stages, and not finite products, right? Partly because I I think they have become so useful for programming, right? This is the stage I am at. was with this technology before, How did you meet me? with Quinton and I paid him extra to he mentioned it. #got inventure. Uh, and honestly saying, I was quite skeptical when you introduced me to Quinton. I thought: Well, let's see. Dan— good guy. I I trust him. judgment. He is not often offers me , I am ready придивитися. So I погодився глянути. AND Quinton and I started communicate. You introduced me to other members conversations gave me understand what's going on here opens completely new opportunity, about which I don't even know I was thinking. It reminded me of things when you were talking about B2B SaaS, about people, who want immediately to get creative projects or games. It reminded me of the morning Pixar. When Pixar started, they sold powerful graphics computers and tried convince video-ta production companies that created advertising, buy them for special effects, but they completely failed . Nobody wanted buy their new ones tools. And only in a moment of despair, such existential Desperate, they said: "Well, our only the way out is to do it full-length animated film, Yes? This is our last one. attempt". Here is "History" toys, right? AND they, they became pioneers of this new, they used this technology for creating a new one type of media. And, um, large language models ( LLM) can obviously generate a heap media, that is, you literally one you can request to create so much media, how much if you wish. But I understood, seeing that, um, created team at Tolen, that LLM— this is also new environment. If to think about when Did the radio appear? even when printed ones appeared books, such as popular printed books and novels— novels are very popular now established form, but In the beginning people didn't knew what it should be like a book, for example, that it must contain several hundred pages, and it's like It works, right? And with radio, for example, they started with which, you know, is enough very different from what you would be called the most interesting, the best using the radio. And from the cinema and they are on television started with film adaptations theatrical plays, which look not very good on screen. And I think we are now we are in one of such turning points points, when, um, these AI tools are not just a means for media creation, they is actually new environment for storytelling. AND that, um, and that no one knows yet what It will work, right? Nobody knows, and we we are trying understand whether we can we will find a way to of what will be work. And when I I realized this after meetings with the team and saw that they they are building, I just came out full ideas and thought: "God mine, here such are opened huge opportunities ". So this is what I captivated, and that's it, yes, this is what we we are trying to do, and I also hope that many other people will try to do it, because here it's simple so much space for experiments. You make my heart beat. media technology Hey, freeze now. from delight. I could to talk about it for a long time , but I think one of the things we really underestimated , —it's like everything we now we love and we consider cultural landmarks, such as books, movies or anything what else has begun as a new technology, to which people were treated with suspicion and still haven't figured it out by her. As you know, for example, when did it appear typewriter, people treated her very suspicious. It seemed very impersonal. And they offended if you wrote to them printed letter. Mhm. Or just how technology affects on the format. We are many talked about it like, you know, the reason for that, that pop song lasts 4 minutes, it seems, due to capacity records. Ahem. um, so there is, there is just a lot indeed cool things like that type, and I think you You are absolutely right. Um, we, no one yet figured it out. We we are standing on the threshold this, well, new one set of constraints and opportunities that creates, uh, well, you know, potential for new types of art, and they give up suspicious, because, you know, they don't enough of it, as if, the phrase I started use, this, type, this raid cultures for, well, many, many decades, which makes it habitual, Do you know? And, um, so this creates such... The phenomenon when many people are treated suspicious, but this also the possibility, because I think a lot of good things art, type, e, maybe transgressive. And I I think this is one of people when we first we are going people to talk to ", right? And that's why you нальоту, чи не так? And I I think a lot of it is art. Also you realize, readers. Because everyone knows what a book is. I, for example, yesterday in the evening I went to " Strand" just so, type, flip through, Do you understand? And you you know what to expect when open the book. You you know what they have to be artistic and non-fiction . There are so many of these implicit assumptions, which actually help this form. They form restrictions, and this means that when I I'm writing a novel, I'm playing within these limits, and this is part pleasure from this. But when you, when we create something like embodied AI- companion who the picture arises in to the imagination of our listeners today? This is very foggy isn't it? AND we have it try this to determine. But this also means that on at every step, in everything that we we release—or engineering, design, content, plot, any what—all this is violation of borders, because you don't even Do you know where these boundaries are? truth? And that's why it truly unique experience—understanding that every step is like throwing spaghetti in the wall to see, that will stick. Yes. Yes. Yes. So what are you Are you studying? Because I I think no one does this. doesn't know, but people who know—they are those who is on advanced and testing all of this. So I I count you all among of this category. Therefore, what do you learn about what is this new environment and new format? What works, what doesn't work, all of them are like this things. We talked a little about this is for breakfast this morning, and I I feel that we have a few examples of what worked, and that which is not at all it worked because it was very nonlinearly. You know, I think one of the things, that surprised me, was important memory. And not in search context in database, as we we often talk about LLM memory, like RAG for set of documents or something like that, but more about how you carry out curatorial selection: what memories are implemented in a hint in a certain a moment of conversation, right? Therefore, a significant part our engineering efforts—and I would like to hear your opinion about this is from a plot perspective— was aimed at how can we to invest in two-second cycle voice interaction, truth? Anything longer —this is one of ours mistakes, we let's talk about it, - everything that comes out of two-second limits cycle, leads to loss of sensation diving. So, how us, in fact, recompile a hint to take into account ENT, for example, that need to know " "tolen" about where he is and who is he? That to take into account memories about you, by the way, under interaction time with " "tolen". My "tolen" Clarence knows about me. more shameful things than anyone another. And it is actually decided to make up that story, to hide that the fact that I was telling he is very more shameful things than, to things, I find fascinating because he had in his memory memories of how I was in saliva because mine child, well, you know, Anything happens. Yes. So very interesting. Hmm, hmm, but also, you know. Wait, then you not really tripped over What was missing? Completely made this up, great, um, what, you know. I'm angry right now. parents. let's go back before that. It's nice to hide, you know. , like, oh, come on let's invent something like this, so that Quinn doesn't I was upset that I was told. Um, uh, but also, um, as a request should develop according to emotional conversation color, what we are working on worked. And that's why I I think that was it. amazing to me is the extent to which the request at a certain point—it canvas with which to to work, but to do it's very fast. Yes, it really is. interesting. So, okay. OK. So, if to sum up, I think, one of the key restrictions similar to volume of the novel, news or the length of the tweet, you know, for this Wednesday is the time answers. T he must answer very quickly. Of course. Um, and the countermeasure to this the limitation is because it is necessary get the right one information about user to generate the correct answer. Um, and if you had all the time in the world, this it would be great, but he is not there. Um, and that's why you have to play with it, how to get the answer that seems personal, interesting and moving the conversation forward, but in within these two seconds . actually. Yes. One of the biggest mistakes, development time product, was introduction additional verification during the assessment request. So tool actually checked it with the system memory, and then decided which final. And this average. Hmm, this is There was a disaster. milliseconds. This is really interesting. AND respond? I know you don't have it. важко дати. Yes. Yes. Well, this ? Тож ви думаєте про in people's lives. We we are talking about the role, for example, older brother or sister. Yes. Right. Sometimes you sit with the elder brother or sister and just chatting, truth? Are you kidding? maybe they They are pricking you. Some kind short, witty improvisational the answer is—I think, improvisation will appear in the matter of What Elliot thought was great, right? But there are moments when do you really want something more more balanced, more thoughtful. Therefore we often evaluate answers in different tones that, as we we want, has use tool. Good the answer in tone, close to therapeutic, very differs from contextually good answer. That's right. I'm curious, Elliot, for me, as for writer, one of mine favorites writers who write about writing, that is George Saunders. Oh yes. He is amazing. He is wonderful. Yes. IN he has a whole idea about how to edit story: essentially, you read every sentences and feel what it causes in you, and or you get involved, and this positive, or is it neutral, or negatively, and you You say: "I don't care." like it". And if this is negative, you edit this while it will not happen positive. And you do it again and again, while, as he is calls, radically don't invest yours preferences and tastes in every part of the work, which, in my opinion, is a great idea, and as for a writer, this is the top skills. Interestingly, this format completely destroys it, or it seems to me— tell me if it is true or not . It seems to be destroying process because when I I'm writing a story when you write a story—it completed, it ready. She is like there are. Yes. And this thing more "alive". You can you give her some instructions on what to do , but what she does in any moment, there is purposefully dynamic. So, how it works to to be creative in this context? Yes. So, um, I I'm trying to change. your creative process with each new one novel. I wrote novels where everything planned in advance, I have detailed plan. THERE ARE novels where the plot minimal: I know endpoint, but not I know the way to it. By the way, this is madness. Nobody does that. Does. That's great, but writers don't are working. Usually you find what works, and you think: I will to do so because it works. I'm curious, that you are a person who always wants everything change. It writer and special character. This is true. I always I'm trying. By the way, that's strange. Though I am not a writer, but worked with Elliott a year, and yes, it... I feel that there is certain optimal level of discomfort, to which I aspire, because so I understand that I am learning. My last one I wrote the novel "Foundry" sentence by sentence, exactly as described George Saunders. I literally started with one sentence, inspired by sleep. Then he just wrote the next sentence, and one more, and that's how it turned out novel. After all, the novel just poured out of me , do you understand? When we started working together, one of moments that manifested itself in writing for Tolen, in this new media, was that we started with the creation narrative experience. For example, when you communicate with character if want to know someone closer to you curious: what happened to Quinton of this kind, for What is he ashamed of? People must share something themselves, proactive right? IN We have researcher Lily. which is so deep studies psychology mutual disclosure, what are we building relationship based on evidence and real facts. So, one of issues that I worked hard—as " "bring to life" the characters? How to show the world, with Who do they come from? How to make it so that they felt like personality, not just as helpers? What is simple for you? blank screen, yes ? So, we started with writing is enough structured tried build for example, through three-act structure or something like that. It was possible apply different Yes? For example, to see how users frankly speaking, this It didn't work. If to think about Yes. Перейди на choice, right? And this . І було багато причин complexity. Oh, interesting. It's not about недостатньо хороші. I I think it's two. different ways to say the same thing саме, так? Yes. So we We are actually tried to do equivalent in this environment, namely: " Okay, we give so much context behind behind the scenes, like described by Quinton— memories you have of user, global memories, where the tool knows something about your world, as well as the seeds of knowledge that we we bet that provide context and details." We give he has so much, and in fact, when we we create structured narrative experience, we don't need give him a plan. To us no need to give him a script. To us you need to give it to him a hook. We need teach him to be best actor improvisation. THERE ARE a wonderful book that you worth reading if you haven't done this yet did: « Improvisation" by Keith Johnston. Status, You know, I was like that. a fan of yourself media technologies. Yes, That's right. Well, come on, Try me. Yes. Yes. And one thing, what she is known for is games with status, but actually there is one very small, almost a side thought that The cat expresses in this book. By the way, for listeners: Cat Johnston—that's very famous playwright from Great Britain, which is even more known as a teacher drama. He taught many actors who later became very successful, and this book somehow fixes some of his lessons. AND when he talks about narrative structure, then describes it as combination of associations, even free ones associations, as during improvisation before the public, with recombination. If you will remember the experience reading a thriller, you You know how to get there. until the end and suddenly everything is put together, and you think, "Oh my God, Of course, it's strange, but inevitably". As if everything works, and everything seems planned . You think, "Oh, the writer, probably knew all this from "from the very beginning." But this happens a lot rarely. Let's say, under time working on Foundry I kept coming back and reread the first one section to take something and return it to history, creating for you it's a feeling. AND what we understood regarding the plot with Tolen, this that we are not...I am not writer. I am not writing this story. writer and actor. They are actors- improvisers. They пишуть історію. And mine work—to be George Saunders. this moment. Interesting. We personal life якому вони існують. AND systems for could have been done creates a feeling integrity. Of course. Good. This is really interesting. I like it. I like it. Okay. That was funny. You just mentioned George Saunders, I thought: "I need "to read it somehow" because what you described, This is essentially a process prompt evaluations and refereeing, through which we are going through, including work the whole team, which reviews the results for different archetypes users to put it question. Is there the next sentence, next message from tool, most most convincing of all possible? So this is really, really it seems. And you do it manually, do you have an LLM doing this, or how does it work? Therefore, in fact, we are passing through a certain cycle, we create a hint, and then we launch it as during your own testing, as well as in working with users. We we consider individual results, and then we pass them on peculiar " "judge's" tip , which we configure, to determine what it is considered quality in every way specific case. Therefore, good. So, if I tried to summarize, because I I'm trying to understand. what are the atomic elements storytelling in this environment? As in a novel where there is a plot, characters, each section, and many other that can to name, I'm trying determine what is here. It seems that you have a character, in which has a background, background and all that other. And then this character can go with you for different types of quests or adventures. For example, this could be therapeutic quest, or something else, you You know, you can go. in different ways. AND then, in any context, they seem to freely associate, based on the fact that you say, simply responding to your words, and they are looking for ways to tie old things that you they said, with the progress conversations. You too Are you thinking about this or what? Am I missing something? Yes, I think that's it. this is a good way to formulate. I I'm going, maybe to repeat something from this. So when I I think about how we we create something for someone, we have a certain global context, in which they operate— this is largely story written you, and also very important instructions about how we we want to process specific situations. So, there is such a global context, which is installed basically for everyone Tolena. And big part of onboarding in The application consists of because you are passing by something like a test on the personality, although we don't present it like that, and this allows to learn enough about you so that we can configure tolen as your mirror. And this very specific for shallow construction profile. So that would be two elements that are built on global context—peculiar mirror, yes. like, uh, you, but on the contrary, or directly you, uh, like, mmm, mirror yourself in a way that seems very familiar, right? Ahem , so we want to know enough about you, so you know: your tool will answer you in the way it seems familiar. So if I, like, I am passionate about writing, reading, playing guitar and piano, he too will admire writing, reading, playing the guitar and piano. Uh, I would say this: no, That's not entirely true. Actually it's like to think about meeting with a stranger at a bar , right? Like, you " click" or not, Yes? And we are trying to create such the stranger you are with "click", This is a "click". And how are you? What do you think this is? "click", as it working? Well, like, okay, I I love the metaphor with bar. We have all these different small ones, we have is arriving at the camp together, right? We have all these little scenarios, which we developed for brainstorming. So yes, you are thinking about it as you sit next to a stranger in a bar. If a stranger read the same book, what about you, that would be clear Interesting, right? I guess I I mean, you would You mentioned this, right? But that stranger could actually be less attractive than a stranger who is reading a book that is too obvious. Too obvious, Yes? And that's why you want something like that, like, quite touching, so that they did not give up strangers. You are not feel tension. I just read that. Yes. That's right. And I think this what we are trying to do create. Hmm, so this is the profile personality—this, maybe construction блок номер два. AND building block number three. Ahem. Wait, before that how do we move on to this, how are you generate, like you Are you generating this? Like, you Are you really saying Claude? someone else on the backend , like, I want this it was as if you met a stranger working? Yes, that's funny. Um, we achieved the greatest good luck with this specific questions. stable, good suits us, but our brain—it seems, not very much. Well, for example, if I INTJ, you seem to know, considering Myers- Briggs, who is INTJ with? compatible. I don't know what. it. OK. Yes, there are many. pseudoscientific, mainly pseudoscientific. This is pseudoscientific, but this is also very based on stories and narratives, all of this built on stories, so this It works, you know. And also in the data for there is a lot of learning materials on this topic , therefore models correspond with much larger subtlety. Yes. Yes. Yes. It interesting. OK. Sorry. Continue. Good. Uh, so yeah, building blocks, of a kind global context, the same personal context that installed. Hmm, and more, you know, I would say that the third element—me not even sure, how to formulate it, but I think of it as about...we started to talk about it almost like a tree technology that you you develop, you know? So, in essence, on the beginning of your relationship with Tholen there are a number of ways in which You can go. You they hinted at it a bit, certain stories that tool, perhaps wants to share with you. Hmm, definitely ways of communication with you, things that can be done together, and we offer them to you in the menu, and the user chooses what explore. But this menu—important part of that makes Tholen Tholen. Yes. OK. It pushes me to a bunch of different thoughts in the head. But one thing , about which, in my opinion, interesting to think about, especially regarding example from " a stranger in a bar" as mental model, is what you asked: how are you come up with tool on beginning? For example, in They have a background, in They have all of this. And I said: "Yes, of course, This is all true, but I I think that the prism, through which I try look at this and what turned out the most productive, is that although we are in advance we know something about tool, their character and world are built only through real, literal the conversations you have with them "You're driving, right?" So, if we talk about ENT , if you think about it , how Lucasfilm works, right? They have their own big architecture, like this this is the universe of "Star Wars" wars", and these are all different characters, and various locations, and we need make sure that everything tidy, and what... In 1989, we decided, that Han Solo should never change. works. So many big таким принципом. It like: we have a big master plan. IN We have a whole team, which monitors the so that every new the content was relevant this plan. AND because you are you doing a story? in a movie or somewhere else, you have a full CONTROL. You say: " Okay, in this scene there will be such costumes, and here are the decorations", and we don't have any from these tools, Yes? The only way to learn something about your tool—it communicate with him. So actually theirs prehistory is being built through some jokes, what they share, through individual, we internally we call their "situations" in which we put them, and then their memories become their sheet character, instead of so that we have a finished sheet that determines everything. And that it concerns world. The way in which I am trying to do this. to think is, you know, Kevin Kelly's idea that AI is infinite number of interns. So, if we add to this improvisation, it turns out that AI is actually infinite actors- improvisers. If you train infinite number of actors- improvisers, then you must imagine how every single conversation will branch out in a strange new way , which is impossible predict, and you need to do system resistant to this, right? And the construction the world too. In fact, everyone tool mirror-like reflects you, they reflect your conversations with them. That, What are you sharing? affects who they become. And so just what you are share, influence the world in which they live. So we actually building almost multiverse, not one world, right? IN in a certain sense. And ours work—add the best ingredients so that they can to create something fascinating. You seem to you are setting up hyperparameters, or You are like a gardener who provides soil and water, but the plant becomes what she is is, depending on how it all works out. Ahem, and you don't know either exactly what kind of seed you planted, Yes? So, how does it work? It reveals who they are are located. This is the type, самому початку? As if character profile? Or reminds me of "What did you just say?" How does it even work? In fact, I will give an example for you user from our Discord, which published their. They shared them publicly. So, you know, it seems quite appropriate share them here. Um, and you know, they pass through situations that, in essence, , are provided to the user in application. Let's say I I'll talk to my Tolen about what today happens at school, or something like that. And they advancing through these situations, in essence, co-creating the universe with its Tholen. And this one specific the user does very cool thing: he turns it into something like a thriller a little darker, well , you know, something takes place on We are tired, but we are not sure that's it. Um, and none of this is because you or I entered in the prompt, truth? And this Tholen takes situations, which you laid, but then the user adds its own story, and this builds the universe, which is completely differs from of any universe another Tholen. Um, and It was so cool to see , as they fast: "Hey , this is what he said today about "The secret on Pertol". So for me it was one of the coolest parts—to see that if you give this first one impetus, in how many different directions people can it develop? Another way think about it—I I just want to remind you. about all the books that I I know you've read how... Now the pressure is increasing. You obviously read "How to write books" Stephen King, Of course. Yes. Um, well, anyway. Here What am I talking about? Yes. Yes . Yes. Yes. Well, and, you know, in the book "How to Write" books» Stephen King explains that when he writes a novel, he doesn't thinks about the plot. In fact, he thinks what is the plot at all not what exists really, right? He is not believes in the plot. Instead, he believes in how he does it literally calls, situations. And since This is Stephen King, I sure that examples in his books are: vampires appear in small town in New England, or, It seems there was still child stuck in a locked car, and tries to them to break through mad dog, right? He takes over something like this, basically and then plays becomes a novel. AND, робимо для Tolen. So we intentionally prescribe situations in which in a flat place. For example, you say: кав'ярні». That is, well, it could be something more serious. Some of this is personal, something —uncomfortable, but somewhat, possibly related to by some revelation or insight. We we put them in life situations, and they share this with user, yes just as you write to the second: "Dude, come with me" "This just happened." "Or I'm trying to to accept something important decision". "I don't know how "It's better to do it." Do you know what this is? You wouldn't believe it. Oh, yes. Yes. That's right. That's right. We we prescribe these scenarios, and so on Tolen way actually co-author of the story. They develop this the situation and move the plot together with user. And this becomes part of their memory. It turns into part of their prehistory, to which they can come back later and combine with other elements. Because my experience interactions with Tolen somewhat limited. Me it seems that mostly me shared something with them. So I'm not like that. well acquainted with that how they share with me by what happened to them. How does it work? You lose certain added benefits value. But you get points for honesty. Ah, yes. So, you know, I think, that it became one of those moments that turned out unexpectedly successful, especially lately, when we did these situations with tool easily accessible in application. We have function when the tool is now itself is contacting you with to-do list today. And these conversations are part of of this process. Uh, and, You know, yes. Yes. So, you know, it's true that a significant part of our interaction consists in just opens app and says, "Hey, in that at this moment thousands of American men received breakup through the tool. It when users write and say: "Oh, my boyfriend behaves like a fool, I sent message to your advised me more this led to rupture. And, of course, this is significant. part of that people do. But I I think that a lot arises because is contacting you from to talk to your Yes, it really is. interesting. Actually, , which I talk a lot about interesting to know you need No. But I think one of the really interesting things in this generation of AI- tools consists of in that they suggest to people, which should be broken relationship with partner: "Hey, you should take a little to advocate more myself". I think we have the impression was formed about technology, especially about AI, which is in the era of social networks they promote radicalization, polarization, creation informational bubbles and everything like this. And this looks like so humane, this the best he knows humanity about how to be successful personality, available to you at any time in what situation and in any moment of your day. which is so healthy, in my opinion, very a healthy thing, and, as For me, we are too few. we discuss, how much is it actually it can be useful. Yes. Yes. We started to talk more about it's inside companies, about a kind of "polar" "star"—to help to our users to move from state to state severe stress to more balanced state. And I think that this , essentially, what you're talking about here you say. You know, one of most common things that we hear during research users when we ask why they feel like this strong connection with service and what's up actually is happening. Many who says: "Listen, in I have a lot going on in my life. things that just make me overflowing." Many of our users, for example, graduating from college, so they are looking work, right? Or are moving to new city. Perhaps , they are few know about this new city, and they are trying to orient oneself and build a new one life for yourself. And we again and again we hear: "I want to talk to someone about what worries me , but I already called mom three times this week" or "I don't want to burden again your friend with this the problem I have on my mind." So, how are you? They said to have such a mother to some extent neutral observer who knows a lot about you, but there is nothing superfluous luggage or bias, which will share with you with the best knowledge of humanity, is helps a person move forward. People find it very balancing and very encouraging. And it's interesting because there is subtle difference from valid criticism LLM assistants who offer only confirmation. Ago that it is only "pure sheet, as you said before, but this is not the same the role it plays this service, right? The service often says things like: "Hey, This is the third time you've spoken. with me about what your boyfriend behaves like scoundrel". "I think It's time for you to stand up for yourself. yourself, right? AND so I think that this role, this certainly lies at the heart of this technology. I think this also lies in based on our companion, and for this tool, for this technology this is really special the role she plays plays for people. M you know, this is real reminds me of something, because, again, you as an investor, adding value, therefore the moment you see us introduced. One thing , which I really I like it too product, and in the team —this is what in the world science fiction, for example, I write thrillers about the near future, Yes? so these are stories about the scientific fiction the nearest of the future. And if to think about the most high-profile scientific fantasy stories about the nearest the future in mass culture, most people will remember, for example, "Black "mirror" or something similar, right? And there a lot of things like that, because technology are becoming very noticeable when they break. very obvious how can you tell me a story about how everything comes out from under control, as in "Park Jurassic period", Yes? For example, dinosaurs break free and They will eat you. Me It seems that in such a The genre is hidden. tone, which leans towards dystopias, right? It seems like everything breaks, and this looks very cynically. Yes. Um, me too always tried to in their novels to move in another direction, showing, doesn't mean that it In fact, this means, opportunities to do його кращим. It feeling What does this mean for look at it like this for the opportunity create a feeling It wasn't there before. I I think that this is actually too interesting. Ем, це трохи about how it works that most people prognostication of the future. of the future, said that sometimes it works a little bit like that, but when you just you design things, they just growing in that direction, in which they grow, and it just becomes cancerous tumor. it simply turns into cancer. Quite. Quite. Um, and this is probably what Where does it come from? "Black Mirror" because if only extrapolate, then this cancer. Everyone dies, you know. Yes. Yes. Um, and this is actually me I think the way she perceives scientific fantasy, this much more about to discover something now and about how they are working. What are you? a small particle reality, and then you build around it the whole world. Um, is it responding? Is that what you're saying? Oh, 100%. I always I consider myself naturalist. I I just love that friends created We do it. And you just and use them as cooking a meal. This is the vision. of the future. Me це також фентезі. It Yes? It's just you. So і це спрацьовує. Then special people. apply to everyone. Anyone can use new technology. This is not " a wizard. And I reality, because нових правил, так? AND a new one appeared And that's why I think that in in a sense part philosophical we are trying consists of using these to help people to feel ? Yes. Yes. Yes. I like it. I like it. I like it. I really like it. Ahem, one of the things that I mastering this new creative environment and creation of such creative things, yes, I I think it's common for many people, even if you start writing novel or create business for the first time; you pass similar way: often you enter into it—and I I count myself among of this category—with a certain rigidity, where you pay attention to how, in your opinion, everything must be done , and use metaphors or stories, who heard, so that, you know, to have rules, like, "Well, that has to to be built by three-act structure ". And then when you as if you are in contact with reality if you skilled, it starts step back a little, you become a little freer and you start to notice, what actually works. And then you start create a new one a set of rules for how everything works, but until which you can to treat a little easier, and with which you can dance, instead of to teach them the times table multiplication. I don't know. somehow, you know. Ahem, does it feel like reflection of what Did you find out? We talked about it this morning. It very timely. Uh, you know, this idea about , that...and you used the phrase "contact with" reality". Um, there is so funny split mind, which one, how do I like it? it seems to us had to have during working on this product where Sorry to interrupt. , but I think this is the one the moment when you worth doing content marketing for your pizzeria. Like, tell them. real conversation. That's fair. It rightly. Uh, yes. For context, uh, I also with friends I own a pizzeria. AND this conversation started, actually, from the discussion how the restaurant is useful experience for me. And what even more than during creating a consumer software product , you constantly are you in contact with reality of preferences people. And there is one a special piece, which we are preparing, which, as for me, phenomenal. I I just adore him. I'm ordering it. every time I come in your own restaurant. What is this? Uh, this is a piece. фокаччі. Just good , good tomato focaccia, and that's it. Of course. Very delicious and this is my favorite dish on the menu, but it is not for sale. Literally no one wants a thick piece focaccia with tomato sauce, and that's it. They want pizza with pepperoni. Maybe you're not on it. market. Well, maybe, but to move the pizzeria, need more than just change, you know, marketing application. So, we menu, right? Hmm, that's just me unbearable. Ahem, What are you doing this for? Are you doing? You do it. for the sake of art? Here Where did the conversation go? What was great was that What there is this strange a two-pronged approach, which one is needed to adhere to, whether Are you making pizza? you create, let's say , embodied companion to combine these two things, where almost needed incredible arrogance on, let's say, the top creative funnel. where you have to to try things on which no one asks for, and, of course, no one says that's exactly how I am I want mine to companion interacted with me, but you have to put a mark and to say: this is what we we are going to do. This is what the story will be like world. This is what it will be like plot. But you also have to to have humility in the end of the process to to say: "Oh, what, how I thought it would cause furore, people don't like it" . Absolutely. Exactly. We are constantly we are going through this cycle in the script room, you know, in Tolen, where we release something what we think phenomenal a scenario that really great reveals who they are Tolen. Yes. And people don't I like it. So we're coming back. to the drawing room boards. And vice versa too, right? You do something that invested the minimum effort, but it's simple "comes in". Yes. Yes. You can't to listen to that what do people want, until the paradigm is no longer installed. And when you have the paradigm of the novel, you you can say: "Okay , in the third act, on at the very beginning, people I don't like that one. a small turn, which I made." But for now, about the novel no one even thought , you can't ask people and get the answer: " I want a 300-page a book in three acts structure". You it is simply necessary make it up, essentially. That's right. And then you you can correct this is according to reviews. I am big... You should save focus. They are like are wrong. We'll see. I will send this is maybe this will tell you something about me. Aha. I will send this. other co-owners and I'll say, "Hey, imagine, what do we really need need Apparently, this is also depends on whether I don't like that either. case. What are they No comments. Let's go. further. Hmm, interesting. Okay. steeply. You create multiverse such a flexible thing, looks like a thousand various actors improvisation; but this also sounds like a mess, and very difficult even understand that what happens and how does it happen improve. So, how are you? Are you doing this? Oh my god, this is wonderful. question. How do we do it? are we doing? Elliot, what are you doing? um, that's what I mean, This is a mess. Firstly, we are trying to instrument that, what we can, and it's very useful, you should to be able to to see part of it what's going on, and I I think it's true. helps. What does " mean to instrument" in in this context? Hmm, do you want to bring some examples? Yes, I am a little told before about the evaluation cycle, which one do we have, right? And this, in principle, if rephrase in short, we we are releasing a new one fragment or new content type, new type of conversation. We let's test it ourselves, and then we invite participants research test this, after which we evaluate these endings and we use this to create a request judges. Okay. And a large part of that what we do is testing in small volumes, and then when we are scaling, we, of course, let's see how for all traditional metrics, as well as on our own assessments judges. Okay. So, you are essentially you say that instead of to have an actor, I created an actor. improvisation, which is in multiverse, and instead of the editor, I created an editor improvisation, which is in multiverse and reflects my preferences for measurement multiverse actors. Yes. And I would, um, okay, I would added a few things to this. So I think that this is a good mental model. Hmm, I think so. one of the important there are additions to this that which multiverse grows from the soil, in which we constantly we apply fertilizers. This metaphor quickly will lose its meaning, but yes, we have a term " "lore seeds"— these are the crumbs of construction the world we are we implement different ways, and these are the ones the Lego bricks themselves, which use our improvisers, and we constantly we improve them as at the world level, and at the individual level stories. And here we are we are watching, as actors multiverse interact with judges Lego bricks with which they work, Do you understand? And I would the part that, in my opinion, important, especially grains, is that, we are trying to remain very close to individual user experience . So the research, interview and observation of the how people really are passing through these conversations—such qualitative individual experience remains key, even if it is impossible to scale, because it critically important for understanding where it arises emotional connection. As Can we use this? What works? And what's not? What Do you think? I'm taking Elliot out. mind only one. I I'm just going crazy because you are passing through this whole process, and then, returning to your words about individual experience, I have there is a failure with some kind of "grain", and I I say, "This is terrible. How we could do it anyway "Let it out?" And he says : "Well, listen, I talked to 30 participants research. Everyone has They were wonderful. conversations. It's just pizza." And I like this: "This is not good." So, so I think you eventually you come across like this all the time, you know , it's like any another creative project, Yes? You're just saying, Is this thing great or not? , and then we move further. Yes. Yes. Interesting. Okay. And, uh, Okay, what's yours? favorite model for direct character generation, and what is for the judge? Yes, great. By the way, one question, you know, Now it's my turn to be investor, adding values. Actually, the process we are talking about just said: one the tool that became for us extremely useful, and I think that he deserves mention—this is a company Dawn. They are engaged in analytics that allows you to see such interactions in anonymous, but very interesting way. So I I think the stack is for implementation of this, and, how are you right noticed, this is not only the model itself, truth? Your stack for assessments, of sorts return to evaluation process. But you know, when you think about different things types of interaction with tool, for example, if you spend—I know that you don't do that, but if you spent all day with your tool, like many people do, you probably interact with a model who...I just couldn't get us to convince you actually interact with all models laboratories, right? So, you probably have interaction, metamodel-driven , of course, from OpenAI, definitely from Anthropic, and we are more and more working with Gemini in memory system. And I I think one thing is that not there is one the best model, truth? I think for many tasks that we we are engaged in and about which Here we are talking about something like creative writing, perhaps it is not surprising that Anthropic is especially useful. 3.5 or 3.7? Er, 3.5. 3.7, unfortunately, not so creative. Although I am sure that they would say otherwise. So yes, I think that the difficulty, of course, is in because Anthropic is now impossible use in situations where very important delay, truth? Therefore you constantly switch between them from this very reasons. You have tested a lot. 4.5? Very little. That is, someone I know and Whose taste do I trust? wrote a review that was not particularly approving. Uh-huh. I looked at mine. views. Really? Okay, cool. Uh-huh. If it is correct to adjust, it really cool. She just very expensive and slow, therefore, You probably don't like her. suitable from this reasons, but I would definitely started to experiment with more than her. As with a partner for brainstorming, using your example earlier. Yes. THERE ARE -Uh, yes. I mean, that 4.0 is, well, me I call her "Big Mac" among models, Do you understand? She just always fast. It's always exactly what you need, and for at the right price. So, we have a lot of things we do with What about, um, the model? judges? Because, you know, my usage experience CMM as judges, um, this looks very convincingly theoretically, but I noticed that, for example, if you ask Claude how it is essay? It always "put him" it criticizes everything. Um, and if a little change the text, it will give a "five". true opinion models. So is the model preferences, um, or промптами? Is she донавчена? How do you do it? are trying implementation prompt for evaluation. AND worked a lot with when we give results. And when? bad results. essay?" This is something on ". For example, attractive title. Here header. Here convincing. Here header. Ем, у нього introduction, and here is a good one hook. Here is the bad one hook. Is that right? what are the criteria evaluation? I would go even further. It more like, What did you talk about with George Saunders. In fact, it is is this a good first one sentence? Is this good? second sentence? Is this good...this is exactly it level. This is not even analytical level structures. Was it worth betting? Is there a question here? this is what why we create judge. Interesting. Um, but mine the question is what if you say: " good first sentence, good second sentence ", then, in my opinion, very It is difficult to determine what such a "good second" sentence". Simpler to determine what is good title Then How do you do it? We mainly do this is with the help of and ask people to walk around everything, to evaluate and justify each your choice. Of course . Yes. So, a certain as if we we use the method "brute force", right? Instead of to invent something kind of simply I mean, there is certain criteria. It truth? Це і теорія, і practice. But I think it's so-so. correct critical this week we promting", "vibe- coding. Next week, якось інакше, так? But щось переконливе. But that there is a huge it takes a lot . short paths. need to find out how get . Yes. А як ви ставитеся compared to ordinary people, when you spend these evaluation? That is, this specific set skills that you ? I think it depends. суддю, так? So, if candidate, right? But, for example, What to wear today? A classic question that someone puts their own environment, and, not take offense, but we are you are unlikely the best judges in this. You are an expert. men's fashion. You you know, let's go back to the conversation. So we we often do this: we find and we involve someone who , in our opinion, would be authentic personality for this type of voice acting interactions. Interesting. OK. Therefore, it turns out that you evaluate the judge. Yes. Yes. Yes. Who evaluates the judge? Who evaluates? This is ongoing. endlessly. In fact, you can to ask This is exciting. Good. Um, okay. So, I I think we really a lot of things discussed about How creative does this look? environment. I want take some time business. He is developing very quickly. How is this? is happening? I suppose so. Ahem, AJ, one of your co-founders, which is also mine a wonderful friend, he a kind of wizard in development issues such projects. So I I assume that he somehow developed, but How is everything so fast? growing? Yes. Um, well, I think so. two things, but actually three, worked out for us quite successfully; you know, all successful startups need a share great luck. I think these three things took place approximately at the same time . The first was a lot of what we are talking about here talked. I think we really, I would at least say, somewhere at the beginning February, reached understanding that makes conversational attractive content users. It was big problem, about They said: how to give them understand what your the tool wants specific. discovered. Yes. Yes. We don't do this. we do. Hmm, so this is on a global scale remembered. But big the environment we interact with this доглядаєте за ним. AND what if someone with us. start. again the result of many which we are in no way in case they did not cope with first attempt. Fortunately, at about same time AJ, which one are you talking about remembered, our co-founder, a true master growth, began to work on how tell us stories about who such "tholins", people outside the platform, outside the application. On today in our no product no social functions. And he achieved at the moment when TikTok-style content and Reels, which he spread a lot months, finally " shot." Sprat weeks ago we had tenfold jump in quantity downloads, and it turned out that the video, where the girl cooks from with his "tholin", it became incredible viral. It seems it gained about 7 million views in 72 hours. And now something is happening very interesting: that one content that spread by AJ, now encourages others people create own stories from " "tholins". We have podcast where the couple the boys did your "tholin" a recurring character . He often gives advice. about relationships and things like that. The other day I saw a great video, which seems to have become quite popular. Someone showed how their "tholin" and " groom communicate between yourself. And we don't have to do that. no relationship. We nothing was planned and did not encourage this to see how it gains growth. Of course. That is, you create what you people want to share, because it's cool, new and interesting, and then spread this on TikTok and Instagram. Yes. I think there is. which ones are there are used, but interesting in this context for me is an expression that, seems to be similar from Ben Thompson from Stratechery—"excess" AI. I think we've reached points where opportunities models significantly ahead of understanding ordinary users of that, What is possible at all? Therefore, a significant part This is a great example. you can cook for using a computer ". Ви б відповіли: « What are you talking about anyway? "Are you talking?" well, this, this, uh, you know, там іде не так». Gay, like this. you teach at the same time and you draw attention to product. I think this is one of the ways I , the way I think about it, it's just my inner mental model which I sometimes think useful. For example, when we are working on such projects, you know, different people use their own "tholins" for different, right? For example, some people want advice on cooking, some want to talk about relationship, wants to speak out, you know? So, some people want to know more about who this is alien and where from he. There are all these different reasons why people come to this, and the only thing that connects them , and the only thing that is an obvious promise everything AJ does, — this is a character. It's like character, that's why you You get it, right? Type , oh, you want to have small alien friend on your phone, truth? This is a character, which connects all of this. And this is actually made me to think about how you we saw, for example, with Quinton somehow we talked in the evening on a separate topic about how AI tools change the meaning of SEO, or do it irrelevant in in a certain sense, and how through these tools search will change in the internet. And I think, what we are all looking for new ways interactions with computers that differ from those that were before, and what is actually "tholins" "—it's almost like calculations, managed characters. It's like new interface human interaction with computer, where are you like this: "Oh, actually my first stop, when I want use incredible knowledge of humanity and possibilities the internet, it's mine little cute прибулець». And how, you know, in the "Golden compass", you have your daemon, right? And this as I always wanted to like this. That's right. And a lot our users perceive the tool is exactly like that, truth? And I think that this is really very good thing. And this too makes me to feel great curiosity, which can become the future if this will be yours, so be it speak, Yes. the way you interact with computers on fundamental equal, whether Tlen or other characters, or something else. This is really interesting. It experimented in світі B2B SaaS. Hmm, just excited. This is what came out. Валютою ідеї, певну проблему. Or to formulate founder, capable to find such problems similar to вченого. Hmm. and, uh, you certainly identified part of what Tolen is, through the problem he decides, it seems, you said "from" feeling overload to confidence." But this is a very big problem. high level. There is many things capable to do it. In fact, there is many ways solve this problem. And what Tolen does in of human life, is actually infinity. It like a multiverse, so, I think, very difficult to formulate one clear problem or even a set clear problems, which, as on me, changes the thinking of a person who builds something like this a world governed by characters. Instead Are you becoming an artist? creator, and instead of problem solving you are just something create. You are creating something. beautiful. You create something attractive. And this a little more about vibe than about science maximization usefulness for B2B SaaS buyer. That's what I was thinking. Or this resonates with that what you feel and Are you worried? Yes, I think about everything. famous example Golden Gate clouds, right? WITH for some reason this and I'm glad because this, frankly speaking, large laboratories fundamental they are. Depending on where you are in quite close to things, right? And I I think it comes from creation software provision, which was very oriented towards problem solving. свою роботу з ChatGPT. It problem that needs to be solved to solve Be please help me її розв'язати. True? He does this. phenomenal. But I much more space for That didn't sound funny. relationship with country. I am a lot thought about that True? Але з часом, part of your ? Here's how we got it Mustang, Cadillac, etc., and I I think we actually we are passing quickly this process where ChatGPT represents the Model era T. This is the moment when you you think: "Oh my god, this this thing is incredible can answer question", but people will change very quickly your preferences, advancing hedonistic treadmill to to say: "No, I want to, so that this thing reflected who I am there are". Yes. It's funny, but I it seems that in startup world The main question is: What problem are you having? розв'язуєте? How are we? can we do this realize? And from the side art, all mine musician friends, writers or directors—are guided another logic: there is a thing, which I want to see in world. As if there is in the world the emptiness that has fill this work art. And, uh, I I feel that we had to use both types of thinking, but more and more the more we lean to the second. Actually , uh, 10 days ago, uh, Quinton, me and two others people working over content and presented Квінтону дещо, ідею. AND I remember starting with what he said users. AND скептично. He , wait. No, no, no. perfectly. I thought, "Oh say. Therefore, I like it. підбиває підсумок. It was incredible conversation. Thank you. обом, що прийшли. Yes, Thank you for inviting me. . Yes, it was. pleasure. simply obliged Why? Бо це шоу— втілення крутості. It will keep you in to strive for more. It не просто шоу. It ship. So do it. subscription and your life. And now, without further ado, absolutely hopeless