18VC Podcast Ep # 8 From a War Zone to Venture Capital — The "Chaos Native Strategist" Tigran Ghukasyan

18VC Podcast Ep # 9 With Model OS Founder Fernando Jia: The Doom of the “Rent Model” Era

From an investor's seat at Y Combinator China (Miracle Plus), Fernando watched AI startups get absorbed by the very models they were built on. Now he's building Model OS so they can own their intelligence instead of renting it. On Episode 9 of 18VC, we sit down with Fernando, founder & CEO of Model OS (Intelligence Cubed) — the "survival infrastructure" that turns AI apps from prompt wrappers into companies that own their own models, evals, routing, and feedback loop. Highlights & Key Insights: * Why he calls most AI apps "prompt wrappers" — and the 3 forces quietly driving a wrapper "extinction event" * The exact day Model OS was born: Jan 20, 2025, when DeepSeek trained a frontier-level model for $6M while OpenAI spent tens of billions * "Turnitin on chain" — how watermarking model weights pays the original creator a cut every time someone fine-tunes their model * When a founder actually needs fine-tuning vs. just RAG — and the foundation-model mistake almost everyone makes first * His contrarian advice to young founders itching to drop out (graduate faster instead) Timestamps: (00:00) Cold open (teaser) (00:42) Welcome & intro (02:09) From IB & YC China to founder (02:42) Why build infra, not another wrapper (05:46) What investing taught him (10:01) The DeepSeek moment (12:11) Why founders couldn't fine-tune before (14:14) The "wrapper extinction event" (19:00) Inside the product (22:51) The CMU compute moat (25:47) Watermarking models on-chain (30:42) RAG vs. full fine-tuning (34:53) The 3-layer router (37:34) Competitors & 2.5M users (41:01) How Model OS makes money (45:44) What data is worth fine-tuning on (49:02) Building a research-heavy team (51:50) "Change maker" (52:39) Milestones: IPOs & the AI bubble (54:48) Don't drop out — do this instead (59:31) Wrap Subscribe for the earliest signal on the founders worth watching. — Connect with Fernando — LinkedIn: https://www.linkedin.com/in/fernando-j/ [https://www.linkedin.com/in/fernando-j/] — Model OS (Intelligence Cubed) — Website: https://modelos.technology/ [https://modelos.technology/] — Connect with 18VC — 18VC Website: https://www.18-vc.com/ [https://www.18-vc.com/] 18VC Podcast on Spotify: https://open.spotify.com/show/6b3YoAMLjZLziwqHWy3EXb [https://open.spotify.com/show/6b3YoAMLjZLziwqHWy3EXb] 18VC Podcast on YouTube: https://youtube.com/@18vcpodcast?si=l69vI6hHwLgIvw4N [https://youtube.com/@18vcpodcast?si=l69vI6hHwLgIvw4N] 18VC Instagram: https://www.instagram.com/18venturecapital/ [https://www.instagram.com/18venturecapital/] 18VC LinkedIn: https://www.linkedin.com/company/eighteenvc/ [https://www.linkedin.com/company/eighteenvc/] 18VC X: https://x.com/Official18VC [https://x.com/Official18VC]

18VC Podcast Ep # 8 From a War Zone to Venture Capital — The "Chaos Native Strategist" Tigran Ghukasyan

He grew up in Artsakh — an unrecognized state he calls a real-world Neverland — fought in the 2020 war, then pulled off a "physically impossible" transfer to the US without losing a single semester. Now Tigran Ghukasyan is building CyNet at USC — and fresh off an exchange semester in Tokyo with Plug and Play Japan, he makes the case for why this could be the age of Japan. On Episode 8 of 18VC, we sit down with Tigran Ghukasyan — founder of CyNet at USC, co-founder at Spheroid, with experience across Plug and Play US, Armenia, and Japan — on failed first startups, pivoting a metaverse company into an XR-native engine, and why the bottleneck of AI is now physical. Highlights & Key Insights: * Why his first startup — repatriating stranded cars through embassies during COVID — failed in every possible way, and what it taught him about the death of middleman arbitrage * The funding trap that kills startups and the board-of-directors trap that kills corporate ideas — and why VC is the only environment that escapes both * The genetics of risk: ~30% of Americans are built to take risks. In Japan, it's less than 2% * How Spheroid pivoted from metaverse to the first XR-native engine — by building momentum with developers instead of pitching decision-makers * US is relentless, Europe chases government funding, Japan has kaizen but no commercialization — and why that's about to flip in Japan's favor * CyNet's thesis: most accelerators disrespect founders' time and build walls; demolish them and bet on "the power of handshake" * His answer to the AI "permanent underclass" debate, and why humans will live in between AI skills * The post-war decision everyone called impossible — and the encircled-commander mindset he wants every uncertain founder to steal Timestamps: (00:00) Highlight reel (00:47) Cold open & meet Tigran (02:13) "Chaos native strategist" (04:57) Growing up in Artsakh, a real-world Neverland (10:29) A first startup that failed in every way (14:05) Startups, corporates & why VC (18:07) The founder–investor loop (20:34) Culture & the genetics of risk-taking (23:20) Spheroid: pivoting metaverse into an XR engine (31:32) Plug and Play: US vs Europe vs Japan (36:36) Quantum & the age of Japan (39:40) Touching the elephant: a vision of the future (44:52) CyNet & the power of handshake (51:31) AI, status & the place of humans (57:34) The "impossible" transfer to the US (1:00:01) Own the chaos: advice for founders Subscribe for the earliest signal on the founders worth watching. — Connect with Tigran Ghukasyan — LinkedIn: https://www.linkedin.com/in/ghukasyan/ [https://www.linkedin.com/in/ghukasyan/] — CyNet — Website: https://courtyardnetworks.com/ [https://courtyardnetworks.com/] — Connect with 18VC — 18VC Website: https://www.18-vc.com/ [https://www.18-vc.com/] 18VC Podcast on Spotify: https://open.spotify.com/show/6b3YoAMLjZLziwqHWy3EXb [https://open.spotify.com/show/6b3YoAMLjZLziwqHWy3EXb] 18VC Podcast on YouTube: https://youtube.com/@18vcpodcast?si=l69vI6hHwLgIvw4N [https://youtube.com/@18vcpodcast?si=l69vI6hHwLgIvw4N] 18VC LinkedIn: https://www.linkedin.com/company/eighteenvc/ [https://www.linkedin.com/company/eighteenvc/] 18VC Instagram: https://www.instagram.com/18venturecapital/ [https://www.instagram.com/18venturecapital/] 18VC X: https://x.com/Official18VC [https://x.com/Official18VC]

18VC Podcast Ep # 7 Building Invisible AI with Bubbl Founder Tony Xiong: The AI Assistant With No App — It Just Texts You

Tony Xiong has been building since he was 15 — first a portable drive 10x faster than anything on the market, then a camera he launched into the stratosphere for Sony. Now he's building Bubbl: an AI assistant that lives in your texts and pulls what actually matters out of your message chaos. No app to install — you just text it. On Episode 7 of 18VC, we sit down with Tony Xiong, founder & CEO of Bubbl, on why the real problem was never information overload, how his Action Context Engine learns who you are, and the on-campus growth stunt that taught him a hard lesson about virality. Highlights & Key Insights: - Why the real problem was never information overload — it's knowing which 100 of your 1,000 daily messages actually matter - The "invisible AI" thesis: why the best productivity tool has no app and no interface — you just text it - Inside the Action Context Engine (ACE): how your personal context, relationship graph, and messaging style decide what's worth interrupting you for - The numbers — 47% day-7 retention, 260 beta testers, and 60,000+ messages sent to Bubbl in two months - The growth mistake: hundreds of campus signups in a day who never came back — and what it taught him about viral distribution - Why he calls Bubbl "an introvert" — it listens in the background so it can strike at exactly the right moment - How Bubbl reads your Gmail and WhatsApp without storing a single message Timestamps: (00:00) Episode highlights (00:58) Welcome to Episode 7 (01:53) From product designer to founder (02:48) Building since 15 & a stratosphere stunt (06:32) Meeting his co-founder in a group chat (07:22) What Bubbl actually is (08:18) The origin story & 20 signups on the spot (09:50) No app — just text "Hi Bubble" (11:18) The real problem isn't overload (12:58) "Invisible AI" & why SMS (15:03) How Bubbl decides what matters (16:58) Inside the Action Context Engine (20:28) Privacy: what it stores (and doesn't) (21:19) Competitors & "not an iMessage product" (23:45) Traction, retention & the moat (25:47) Go-to-market & the viral-marketing mistake (30:26) Why Bubbl is "an introvert" (32:34) Hiring & a people-first culture (34:38) Decisions & the message you never knew you missed (37:54) Life beyond building Subscribe for the earliest signal on the founders worth watching. — Connect with Tony Xiong — LinkedIn: https://www.linkedin.com/in/haoyu-tony-xiong/ — Bubbl — Website: https://bubblai.com/ Try it: text "Hi Bubbl!" to +1 (949) 283-8690 — Connect with 18VC — 18VC Website: https://www.18-vc.com/ 18VC Podcast on Spotify: https://open.spotify.com/show/6b3YoAMLjZLziwqHWy3EXb 18VC Podcast on YouTube: https://youtube.com/@18vcpodcast?si=l69vI6hHwLgIvw4N 18VC Instagram: https://www.instagram.com/18venturecapital/ 18VC LinkedIn: https://www.linkedin.com/company/eighteenvc/ 18VC X: https://x.com/Official18VC

18VC Podcast Ep # 6 Building the AI Operating System for Apparel Manufacturing with ThreadLabs Co-Founders Rylan Jimenez & Diora Juraboeva

70% of US apparel manufacturers still collect production data manually. Only 16% have real-time visibility across the floor. Most shops still run on WhatsApp threads, Gmail, and the foreman’s memory. Rylan Jimenez and Diora Juraboeva, the co-founders of ThreadLabs, are building the AI intelligence layer for the apparel production industry — not another ERP, but a layer that sits on top of whatever tools a shop already uses, so it works for both a 40-year-old LA cut-and-sew factory with no software and a Chinese factory with an ERP from 20 years ago. In this episode, the hosts Philip and David sit down with Rylan and Diora to unpack how a USC Marshall finance student and a Formula-SAE composites engineer ended up building a manufacturing OS together. We get into how they actually met (month two of college, on Formula SAE), the early ThreadLabs idea that came out of a Google hackathon in 2024, Rylan’s years as a freelance fashion designer, hundreds of hours Rylan spent on factory floors in DTLA, the pivot from a brand-facing AI design tool to a manufacturer-facing ops layer, the macro reshoring tailwind, how the Smart Timelines / Risk Flags / Update Autopilot features actually work, what the Techstars Catalyst program forced them to figure out, and Rylan’s closing advice for student founders — that genuine value doesn’t come from AI, it comes from solving the problem, and that the right first step is to do it without any tech at all. A conversation about staying close to the floor, picking the right wedge, and building in an unsexy, deeply offline industry. — Chapters — (00:00) Highlight reel (00:43) Welcome & how we met Rylan and Diora (02:54) Rylan's path: mech engineering → VC at USC (06:30) What working in VC actually teaches you about building (12:57) Why building beats investing (for now) (15:14) Diora's building journey: F1, USC Racing, Honda, Fisher-Price (17:43) Founding VC Academy together at USC (19:53) How Rylan and Diora actually met (Formula SAE, month two) (23:08) The original ThreadLabs idea: an AI tool for fashion designers (25:48) Rylan's freelance fashion-design years (16–19) (28:07) Where the name "ThreadLabs" came from — and the rebrand coming (29:16) Why now: tariffs, reshoring, and the manufacturing tech gap (32:43) The pivot: from "better ERP" to an intelligence layer on top (35:39) Hundreds of hours on DTLA factory floors — what manufacturers actually want (39:31) What the product does today vs. six months ago (44:21) Onboarding, ERPs, and going white-glove on day one (45:34) Smart Timelines, Risk Flags, and Update Autopilot — how they work (48:23) DTG, screen print, embroidery, cut-and-sew — printing types, not products (50:41) Traction: 5 pilot shops, $50K generated, and where the new product stands (53:27) Scaling beyond LA — international manufacturers and other verticals (56:35) The 3-person founding team (with CTO Dennis Pavlov (01:00:22) What the Techstars Catalyst program actually taught them (01:02:20) One word other than "founder": boring vs. people-orientated (01:05:18) The hardest part of being a founder: scope creep and narrowing down (01:08:47) How being a founder changes the way you live with uncertainty (01:11:26) Mentors, advisors, and Rylan's closing advice for student founders (01:16:08) Outro — Connect with ThreadLabs — Rylan on LinkedIn: https://www.linkedin.com/in/rylan-jimenez/?skipRedirect=true [https://www.linkedin.com/in/rylan-jimenez/?skipRedirect=true] Diora on LinkedIn: https://www.linkedin.com/in/dioraj/?skipRedirect=true [https://www.linkedin.com/in/dioraj/?skipRedirect=true] ThreadLabs on LinkedIn: https://www.linkedin.com/company/thread-labs-sc/ [https://www.linkedin.com/company/thread-labs-sc/] ThreadLabs Website: https://threadlbs.com/landing [https://threadlbs.com/landing] — Connect with 18VC — 18VC Website: https://www.18-vc.com/ [https://www.18-vc.com/] 18VC Podcast on Spotify: https://open.spotify.com/show/6b3YoAMLjZLziwqHWy3EXb [https://open.spotify.com/show/6b3YoAMLjZLziwqHWy3EXb] 18VC Podcast on YouTube: https://youtube.com/@18vcpodcast?si=l69vI6hHwLgIvw4N [https://youtube.com/@18vcpodcast?si=l69vI6hHwLgIvw4N] 18VC Instagram: https://www.instagram.com/18venturecapital/ [https://www.instagram.com/18venturecapital/] 18VC LinkedIn: https://www.linkedin.com/company/eighteenvc/ [https://www.linkedin.com/company/eighteenvc/] 18VC X: https://x.com/Official18VC [https://x.com/Official18VC]

18VC Podcast Ep # 5 "You can't AI your way out of bad taste" — Diana Melencio (GP, XRC Ventures) on building the next $100M consumer brands

"You can't AI your way out of bad taste." Diana Melencio is the General Partner of XRC Ventures' Brand Capital Fund — a NYC-based seed-to-Series A consumer fund whose LPs include some of the largest public retailers, CPG companies, and global mall operators in the world. Before becoming a GP, Diana was a two-time consumer-tech founder (OK My Outfit and Quinn) and spent a decade in equity research and growth equity at Robertson Stephens, Morgan Keegan, and Cramer Rosenthal McGlynn — including time on a Financial Times-ranked #1 Specialty Retail team. In this episode, Diana joins Lucas and Philip from Paris to explain why consumer VC is the most contrarian bet in the market right now — and why her fund's diligence process looks more like growth equity and private equity than traditional venture. In this conversation, you'll hear: * The difference between a good idea, a good business, and a venture-backable business — and why Diana now tells most consumer founders to skip VC entirely * Why XRC "sets up the acquisition from the beginning" — and the Venn diagram method she uses to source brands, starting from Fortune 500 CFO conversations * How XRC built a proprietary AI tool to track newly launched DTC brands via their Shopify checkout pages, and why she still believes consumer sourcing is fundamentally relationship-driven * The two metrics that matter most in her diligence (hint: repeat purchase rate and price-per-fluid-ounce) * What "taste" actually means in a founder — illustrated through the origin stories of two XRC portfolio companies: Homefield (built by a college student screen-printing vintage Indiana sports tees in his dorm) and Naked Sundays (founded by an Australian news correspondent who watched colleagues get skin cancer cut out of their faces) * Why the post-Allbirds, post-Casper reality of consumer IPOs has reshaped how she invests * Diana's honest answer on whether she'd found again — and the lengths she went to keep her first two startups alive (Christmas trees and Airbnb-ing her spare room included) If you're building, backing, or thinking about consumer brands in 2026, this conversation is worth 45 minutes of your week. CHAPTERS (00:00) Highlight reel (00:41) Welcome & introduction (01:36) From Wall Street to fashion: Robertson Stephens, the mentor who told her to leave, and finding 20,000 users in two months (06:17) Why fashion, and the virtual try-on problem she's still bearish on (09:13) Lessons from 2x founding: good idea vs. good business vs. venture-backable business (11:56) Navigating imposter syndrome as a GP (14:45) Inside XRC: two funds, one ecosystem, and the symbiotic deal flow between them (17:08) Biggest success vs. bitterest lesson — and the AI co-founder question (19:09) XRC's edge: C-suite relationships at strategics (21:16) Why consumer VC is the contrarian bet of 2026 (23:42) AI in deal sourcing: XRC's proprietary brand-tracking tool (26:27) The diligence process: thematic, growth-equity-style, and built for strategic exit from day one (31:09) What "taste" means: the Homefield and Naked Sundays origin stories (36:42) What separates a high-potential consumer founder (39:33) The most common fundraising mistakes (41:43) Would she found again? "It was soul-crushing." (42:55) Where to follow Diana (43:35) The childhood dream and the path here More on XRC Ventures and Diana: XRC Ventures — https://www.xrcventures.com [https://www.xrcventures.com] XRC careers (Silicon Valley intern role) — https://www.xrcventures.comhttps://careers.xrcventures.com/companies/xrc-ventures-2/jobs/78500795-2026-analysts-part-time-interns#content [https://careers.xrcventures.com/companies/xrc-ventures-2/jobs/78500795-2026-analysts-part-time-interns#content] The Brand Capital Report (Diana's Substack) — https://brandcapitalfund.substack.com [https://brandcapitalfund.substack.com] Diana on LinkedIn — https://www.linkedin.com/in/dianamelencio/ [https://www.linkedin.com/in/dianamelencio/] XRC Ventures on LinkedIn — https://www.linkedin.com/company/xrcventures [https://www.linkedin.com/company/xrcventures] More on 18VC & 18VC Podcast: 18VC Website: https://www.18-vc.com/ [https://www.18-vc.com/] 18VC Instagram: https://www.instagram.com/18venturecapital/ [https://www.instagram.com/18venturecapital/] 18VC LinkedIn: https://www.linkedin.com/company/eighteenvc/ [https://www.linkedin.com/company/eighteenvc/] 18VC X: https://x.com/Official18VC [https://x.com/Official18VC]