The Luba Show

Today I’m talking to Dylan Fox, founder and CEO of AssemblyAI (https://www.assemblyai.com). 

Dylan taught himself to code, was a machine learning engineer at Cisco, and went through Y Combinator's very first AI batch in 2017, and now AssemblyAI serves over a million developers and 100 million API calls a day with $130M raised from Accel, Insight and 
Y Combinator. 

On October 27, AssemblyAI launches AssemblyAI-TTS-1, a text-to-speech model built for voice agents that completes its voice stack alongside streaming speech-to-text and turn detection. He gets into the five years he waited for a Series A, and why he thinks the last 10% of an AI product is where product market fit gets decided.

We get into:

- Five years from founding to a Series A, in a voice AI market he says has since grown 10x
- Why AssemblyAI went 100% usage-based with no committed contracts, and how that produced around 180% net dollar retention
- The Amazon Echo commercial that pulled him into voice while he      was at Cisco
- Applying to YC a month late with no co-founder, and how partner Daniel Gross vouched for him
- Scaling headcount before the business was ready, and the go-to-market reset that left AssemblyAI at 85 to 90 people

and so much more!

𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒
(00:00) Trailer
(01:16) Dylan Fox, Founder & CEO, AssemblyAI
(01:46) YC's first AI batch and five years to a Series A
(05:49) Granola, Tolan and the voice agents built on AssemblyAI
(07:27) Why the last 10% of an AI product creates product market fit
(09:43) 180% net dollar retention with zero committed contracts
(14:04) Field sales techs earning $20K more commission with voice AI
(17:47) Voice as the primary interface to all hardware and software
(24:42) Teaching himself to code to build a DoorDash for toiletries
(29:17) Rejected by Twilio, hired by Cisco
(33:07) The Amazon Echo that started AssemblyAI
(40:56) Applying to YC a month late with no co-founder
(45:01) How Daniel Gross got him into YC
(1:01:30) Scaling up too early and resetting go-to-market
(1:07:21) Leaning on Accel, Insight and founders from Synthesia and Pinecone
(1:14:19) Why he quit X
(1:15:10) Outro

Follow Dylan
LinkedIn: https://www.linkedin.com/in/dylanbfox/
X: https://x.com/YouveGotFox
Website: https://www.assemblyai.com

Follow Luba
Instagram: https://www.instagram.com/lifeoflubaa/
LinkedIn: https://www.linkedin.com/in/yudasinal/
X: https://x.com/LubaYudasina

Today I walk with Ron Nachum, co-founder and CEO of Sapien, an AI company that helps CFOs understand how their businesses run. Ron dropped out of Harvard to start Sapien with two friends from high school, recently raised from Neo and Ali Partovi at a $180M valution. Sapien works with CFOs at manufacturers and Fortune 500 companies, and now at 22 he's building an engine that reads a company's financial and operational data from the ground up. He gets into the night his co-founder found a $10M mistake hours before a customer's board meeting, and why he picks investors by asking if he'd want them at his wedding.

We get into:

- The spreadsheet error that turned $10M in expected profit into a $2M loss, and how Sapien caught it the morning of the board meeting

- Why he'd rather start a hard company than an easy one when anyone can build anything with AI

- Dropping out of Harvard after Ali Partovi's first check from Neo, and why the risk felt asymmetric

- The wedding test he uses on every investor, and why he chose Neo over Y Combinator

- Juggling 500 times in a row in second grade, and the nurture-only belief behind everything he builds

so much more!

𝐓𝐢𝐦𝐞𝐬𝐭𝐚𝐦𝐩𝐬

(00:00) Trailer
(01:18) Ron Nachum, Co-Founder & CEO, Sapien
(01:35) The $10M error found on board meeting day
(04:53) Telling CFOs twice his age they were wrong
(07:55) How a 19-year-old earns trust from skeptical buyers
(11:53) Why it's better to start a hard company
(13:29) Rubber band bracelets and 3D-printed fidget spinners
(18:32) The professor who told a 14-year-old not to email him
(22:56) Pain tolerance as the top founder trait
(27:16) Being too blunt as a young leader
(35:43) Why he records almost every conversation
(38:31) Dropping out of Harvard after Neo's first check
(45:30) Choosing Neo over Y Combinator and the wedding test
(46:54) From recruiting tools to the brain for CFOs
(54:45) Finishing SOC 2 controls overnight to close a customer
(59:18) Why he wants to build a hotel on the moon
(1:00:45) Outro

Follow Ron
LinkedIn: https://www.linkedin.com/in/ron-nachum/
X: https://x.com/RonNachum
Website: https://sapien.ai/

Follow Luba
Instagram: https://www.instagram.com/lifeoflubaa/
LinkedIn: https://www.linkedin.com/in/yudasinal/
X: https://x.com/LubaYudasina

Today I walk with Quinten Farmer, co-founder of Tolan (https://www.tolans.com/), a personal AI built to be a guide through life. Quinten has sold his last company, Even, to Walmart for $300M, saved millions of Americans billions in payday loans and overdraft fees, and raised $60M from USV and Khosla. He gets into why he has zero interest in an AI that books your flights, and why Tolan takes most of its time from TikTok and television.

We get into:

• Why he has zero interest in an AI that books flights or writes work emails, and the guardian angel idea behind Tolan
• Social penetration theory, and why every Tolan gets its own lore and backstory
• Selling Even to Walmart for $300M after saving Americans billions in payday loans and overdraft fees
• The email to David Tisch that gave him two options: hire me or I go to Afghanistan
• The lunch spent on compostable takeout container margins that pulled him back into tech

so much more!

𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒

(00:00) Trailer

(00:49) Quinten Farmer, Co-Founder, Tolan

(00:58) What Tolan is and what it is not

(02:02) The car accident that showed Tolan's agentic side

(04:47) Anthropic's machine God VS being nice to your toaster

(06:24) Social penetration theory and why every Tolan has a backstory

(09:36) 200 minutes a week and taking share from TikTok

(10:47) Why great friendship gets mistaken for therapy

(14:26) Talking to his Tolan at 2 a.m. in the hospital

(16:32) Why Silicon Valley is obsessed with AI assistants

(21:38) Dropping out and the only time he saw his grandfather cry

(29:35) The David Tisch email: hire me or I go to Afghanistan

(32:22) Selling Even to Walmart for $300M

(37:24) Why shipping a lot with AI is the wrong move

(57:43) How Tolan runs without 996

(1:03:38) What he wants his boys to remember

(1:04:30) Outro

Follow Quinten

LinkedIn: https://www.linkedin.com/in/quintendf

X: https://x.com/quintendf

Tolan - https://www.tolans.com/

Follow Luba

Instagram: https://www.instagram.com/lifeoflubaa/

LinkedIn: https://www.linkedin.com/in/yudasinal/

X: https://x.com/LubaYudasina

Today I walk with Ali Partovi, founder and CEO of Neo and host of a new podcast Confession. Ali co-founded LinkExchange, which Microsoft bought for $265M, co-founded Code.org with his twin brother Hadi, and wrote the first check into Cursor, and now he runs Neo, the firm that backed Kalshi when every top VC passed. He gets into the night he wanted to tell Universal Music Group to f*** off, and the $12M he borrowed against his own account to keep Kalshi open before the 2024 election.

We get into:

- Why he wanted to tell Universal Music Group to f*** off over a Bono video, and the half measure he still regrets

- How Sequoia's Mike Moritz judged hires in ten minutes while three co-founders spent an hour each and had no opinion

- Betting on Alfred Lin as LinkExchange's CFO when Ali thought the idea was idiotic, years before Alfred became co-head of Sequoia

- Wiring Kalshi $12M of borrowed money when no bank would lend $100M two weeks before the 2024 election

- Why Elon Musk stays the face of SpaceX, and why founders shouldn't fight being the face of their company

so much more!

𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒
(00:00) Trailer
(00:55) Ali Partovi, Founder & CEO, Neo
(01:09) Rock climbing as exposure therapy for fear
(06:47) Bono, U2 and the Universal Music threat
(16:00) Selling iLike for $25M after a 10x bigger first exit
(17:44) Nobody cares about your failure as much as you do
(22:51) Why Elon Musk stays the face of SpaceX
(28:50) From MIT daycare to the Iranian revolution
(37:00) The Commodore 64 stuck in Iranian customs
(47:59) Mike Moritz and his 10-minute hiring verdicts
(54:19) Betting on Alfred Lin before Sequoia
(58:05) Spotting Paul Graham and Robert Morris at Viaweb
(1:03:35) Flying campus to campus to start Neo
(1:13:00) The Cursor founders who named their team "the winning team"
(1:17:46) Kalshi's 40 law firm rejections
(1:23:25) Wiring $12M to Kalshi before the 2024 election
(1:33:58) 40 Neo scholars inside OpenAI
(1:41:46) Why Neo is named after The Matrix

Follow Ali
Confession: https://link.mgln.ai/4TnDcL
X: https://x.com/apartovi
LinkedIn: https://www.linkedin.com/in/apartovi

Follow Luba
Instagram: https://www.instagram.com/lifeoflubaa/
LinkedIn: https://www.linkedin.com/in/yudasinal/
X: https://x.com/LubaYudasina

Today I walk with Bryant Chou, co-founder of Webflow and founder of Ploy. Bryant is the former CTO of Webflow, the platform behind roughly one in every hundred websites, built its growth and marketing teams, and went back through Y Combinator 13 years after his first batch, and now he's building Ploy, an AI platform that turns your website into your hardest working employee. He gets into deciding to leave Webflow in a single day, and the grandfather whose last question to him was how many people worked at his company.

We get into:

- Why he left Webflow after 13 years to start over solo at 40, and why he decided in a single day

- Being "scared s***less" of 20-year-old founders, and why he said yes when Garry Tan asked him to do YC again

- Why building a company in 2026 looks nothing like 2013, when he, Vlad, and Sergey had to cover every skill themselves

- Ploy's bet on the 99% of websites that aren't on Webflow

- His grandfather's escape from a wanted list in Taiwan, and the "keep going" that followed him through 800 employees

so much more!

𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒
(00:00) Bryant Chou, Co-Founder of Webflow & Founder of Ploy
(01:46) What running Webflow's marketing team taught a CTO
(03:05) "Slurping" your website into a brand design system
(07:35) Deciding to leave Webflow in a single day
(11:08) Going after the 99% of websites not on Webflow
(12:34) Turning your website into your hardest working employee
(13:41) Scared of 20-year-old founders and doing YC again
(14:30) Building a startup in 2013 VS 2026
(18:04) Why a 40-year-old third-time founder went back to YC
(20:16) Demoing Ploy cold after his son's baseball game
(24:31) His grandfather's escape from a wanted list in Taiwan
(35:41) 996, FOMO, and the San Francisco rat race
(40:10) Relearning founder mode
(50:58) Running three teams through Covid with two kids under two
(58:00) The redemption he's chasing with Ploy

Follow Bryant
LinkedIn: https://www.linkedin.com/in/bryantchou/
X: https://x.com/bryantchou

Follow Luba
Instagram: https://www.instagram.com/lifeoflubaa/
LinkedIn: https://www.linkedin.com/in/yudasinal/
X: https://x.com/LubaYudasina

Today I walk with Adam Delehanty, founder and CEO of Ghost, a leading ghostwriting agency for founders, VCs, and brands. Adam has written culture books for Hinge, Duolingo, and e.l.f. Beauty, advises many prominent founders and VCs on their content, and has hosted a popular writing group in his sunroom for 175 Wednesdays in a row. These days, he's been spending more time stepping out from behind the “ghost” and building his own voice online. 

We get into:

• Why ghostwriting = journalism × therapy
• Growing up the product of Irish writers and anxious Jewish therapists → "This was the only job that made sense”
• Why culture starts to dissipate somewhere between 200 and 1,000 employees, and the role of culture artifacts and lore in preserving aura 
• His method for using AI: the human + Claude + human sandwich, and how to train Claude to edit its own work
• Why founders who only post about their product are that person at that bar everyone runs from
• Why he spent four days in the woods recently with his phone locked in the car, and what on earth happened
• Why words like "partner" move us further from the truth

so much more!

𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒

(00:00) Trailer

(01:17) Adam Delehanty, Founder & CEO, Ghost

(03:07) Why culture breaks between 200 and 1,000 people

(04:28) What it means to become an “internal journalist” for his clients 

(07:06) Hiring a ghostwriter is nothing to be ashamed of!

(08:38) Cutting 12 company values down to the 4 that are real

(09:59) The right reasons to start a company

(13:43) The founder who only talks about his product at a bar

(15:02) Why my LinkedIn posts get three likes

(16:59) Human, Claude, human: how Ghost writes

(19:36) Training Claude to edit its own writing

(20:59) Why AI won't replace the hardest part of his job

(22:13) Writing as a way to figure out who you are, and what your company is

(25:51) Four days in the woods with his phone locked in the car

(29:20) How Williamsburg lost all its aura 

(31:22) Reading Ted Sorensen instead of TikToks about him

(34:00) The subtle hints of particular words: why women call some men “dude”

(35:51) Notes on trendy lingo in tech right now (e.g., “GTM engineer”)

(39:33) The Ghost mantra: “find the language your ideas deserve”

(40:51) What he's still becoming

(41:34) Outro

Follow Adam

LinkedIn: https://www.linkedin.com/in/adamdelehanty/

X: https://x.com/adam__xyz

Follow Luba

Instagram: https://www.instagram.com/lifeoflubaa/

LinkedIn: https://www.linkedin.com/in/yudasinal/

X: https://x.com/LubaYudasina

Today I walk with David Haber, general partner at Andreessen Horowitz. David co-founded Bond Street and sold it to Goldman Sachs, started his career working for the founder of Royalty Pharma and investing at Spark Capital, and helped open a16z's New York office. He gets into why he'd rather back the missionary than the mercenary, and why he believes opportunity lives between fields of expertise.

We get into:

- The Friday night football taco stand he set up in high school, which he says made the vendor millions of dollars over the next 20 years

- Working next to Airbnb co-founder Nathan Blecharczyk as a sophomore intern, and passing on the chance to become Airbnb's first non-technical hire

- How Royalty Pharma got to a $25 billion market cap with about 50 employees, and why it inspired his firm over fund thesis

- Vantage, the AI tool he built from 750 meetings' worth of transcripts that now lives in a16z's Slack

- Why founders spend 95% of their time selling, and why he backs missionaries over mercenaries

so much more!

𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒
(00:00) Trailer
(00:51) David Haber, General Partner, Andreessen Horowitz
(01:14) Why people call him the most connected VC in New York
(02:35) From the Holocaust to Nicaragua to Tijuana
(05:23) The Friday night taco stand that made a vendor millions
(11:18) Sitting next to Airbnb's Nathan Blecharczyk as an intern
(15:40) Royalty Pharma: $25B market cap with about 50 employees
(28:48) The hardest problem in building Bond Street
(37:06) Emailing Goldman Sachs' CFO and ignoring his job description
(39:27) Vantage: the AI tool built on 750 meetings
(46:02) Why founders spend 95% of their time selling
(47:20) Missionary VS mercenary founders
(52:18) Firm over fund, and how a16z runs like a company
(59:40) Taking 275-year-old Sotheby's into NFTs
(1:00:36) Marc Andreessen's pitch: the J.P. Morgan of the information era
(1:10:45) Why he wishes Bond Street had been bigger
(1:12:31) Outro

Follow David
LinkedIn: https://www.linkedin.com/in/haber/
X: https://x.com/dhaber?lang=en

Follow Luba
Instagram: https://www.instagram.com/lifeoflubaa/
LinkedIn: https://www.linkedin.com/in/yudasinal/
X: https://x.com/LubaYudasina

Today I walk with Shishir Mehrotra, CEO of Superhuman, the company formerly known as Grammarly. Shishir co-founded Coda and sold it to Grammarly, built skippable ads at YouTube into a product that went from $0 to $1 billion in 18 months, and sits on the boards of Spotify and Walmart, and now he runs a company with over $700 million in revenue. He talks about why most people should not start a company, and why he gave up ownership of Coda without any startup pride.

We get into:

- Why the right startup idea should feel like a curse, and why he tells students not to keep a list of startup ideas

- The misspelled Vinod Khosla email that made $500,000 land in his account and moved his company to Silicon Valley in a week

- Joining YouTube when it lost money on every single view, and board meetings with Eric Schmidt about shutting it down

- The skip button every salesperson said would kill revenue, and how it went from $0 to $1 billion in 18 months

- Selling Coda to Grammarly, and why he needs a sense of purpose more than a sense of ownership

so much more!

𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒

(00:00) Trailer

(01:07) Shishir Mehrotra, CEO, Superhuman

(01:32) Making the Lego robotics world championships four times with his daughters

(08:59) His dad only paid for majors where the books are 50 years old

(12:18) Why most people should not start a company

(14:47) The Vinod Khosla email that moved $500,000 in one afternoon

(20:15) Joining YouTube when it lost money on every view

(25:50) Board meetings with Eric Schmidt about shutting YouTube down

(35:20) The skip button that went from $0 to $1B in 18 months

(40:30) Susan Wojcicki and the 55/45 creator revenue split

(45:54) Sketching Coda on a notepad by his bed

(53:52) Why he sold Coda to Grammarly

(59:30) Purpose over ownership and zero startup pride

(1:01:45) The bundling theory he built with Daniel Ek at Spotify

(1:04:49) Superhuman's AI agents and 60% of auto-drafted emails sent

(1:09:37) Why AI will make teams bigger

(1:15:32) Outro

Follow Shishir

LinkedIn: https://www.linkedin.com/in/shishirmehrotra/

X: https://x.com/shishirmehrotra

Follow Luba

Instagram: https://www.instagram.com/lifeoflubaa/

LinkedIn: https://www.linkedin.com/in/yudasinal/

X: https://x.com/LubaYudasina

Today I walk with Noor Siddiqui, founder and CEO of Orchid. Noor was a Thiel Fellow at 17, studied CS at Stanford, and has raised $16.5M for Orchid from Vitalik Buterin, Brian Armstrong, Balaji Srinivasan, and Elad Gil, and now she's out to end what she calls the genetic lottery. She gets into why she thinks people won't have sex for babies anymore, and what she'd pick for her own kid if the tech existed today.

We get into:

- Why she calls the genetic lottery "super evil and unfair"

- The 5–10% chance of genetic disease if you conceive the old-fashioned way

- Her most controversial take: people aren't going to have sex for babies anymore

- Why Michael Phelps is only Michael Phelps because of a genetic lottery

- Tetrachromats, 100 million colors, and what "leveling the playing field" actually means

so much more!

Timestamps

(00:00) Trailer
(00:53) Noor Siddiqui, Founder and CEO, Orchid
(01:04) The flashmob dance proposal
(11:15) Applying to the Thiel Fellowship at 17
(24:22) Founders raising $50M and losing $100M
(31:48) Taking advice from the 500 IQ nerd
(35:36) Why genetics shape everything
(51:03) What is a super baby
(54:55) Sequencing 3 billion letters and why the genetic lottery is evil
(1:01:16) People aren't going to have sex for babies
(1:02:43) Michael Phelps and the genetic lottery
(1:07:21) Escaping the code we were stuck with 
(1:20:21) Upgrading all humans would make her cool
(1:22:20) Outro

Follow Noor

LinkedIn: https://www.linkedin.com/in/noorsiddiqui/

X: https://x.com/noor_siddiqui_

Follow Luba

Instagram: https://www.instagram.com/lifeoflubaa/

LinkedIn: https://www.linkedin.com/in/yudasinal/

Today I walk with Gagan Biyani, founder and CEO of the Horowitz Andreessen Academy. Gagan co-founded Udemy, which grew into a multibillion-dollar public company, built Sprig, co-founded Maven, and started his first profitable business at 13, and now he's building a16z's new school for young builders, backed by $42 million. He gets into the $60M company he had to shut down, why his investors backed him again anyway, and why a Harvard dropout means nothing to him.

We get into:

-Starting a debate camp at 13 that ran 50% EBITDA margins, and landing in the principal's office for it

-Sprig: $60M raised, 1,300 employees, a shutdown, and investors who backed his next company anyway

-Why the Academy flips college around, with students building instead of attending class

-The call from Andrew Chen and ten meetings with Marc Andreessen and Ben Horowitz that became the Academy

-Why he tells 18-year-olds that raising VC is a ten-year commitment

so much more!

𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒
(00:00) Trailer
(01:01) Gagan Biyani, Founder & CEO, Horowitz Andreessen Academy
(02:23) A one-year fellowship built around building, not class
(03:55) The three skills that are AI proof
(08:16) Grandparents from Rajasthan who never finished high school
(09:31) Starting a debate camp at 13 with 50% margins
(12:59) Why schools are afraid of student agency
(18:33) Seven jobs, three years, and rarely going to class
(18:46) Why most professors don't care about teaching
(25:41) What Udemy and Maven taught him
(33:32) Raising $60M at Sprig, shutting down, and getting backed again
(38:16) Why raising VC at 18 is a ten-year commitment
(39:21) The call from Andrew Chen that started the Academy
(41:18) Why the best builders will pick a16z over MIT or Harvard
(43:17) 50 founding students, tuition-free
(45:46) Why a Harvard dropout means nothing to him
(51:42) Outro

Follow Gagan
LinkedIn: https://www.linkedin.com/in/gaganbiyani/
X: https://x.com/gaganbiyani

Follow Luba
Instagram: https://www.instagram.com/lifeoflubaa/
LinkedIn: https://www.linkedin.com/in/yudasinal/