A VC who burned $3M on the wrong market — and how the same startup came back as a $50x bet.
🎞 RYA-ZVE1-1143.MP4 · ~96m🎞 RYA-ZVE1-1144.MP4 · ~24m🎙 Igor Kaloshin — guest; CEO & co-founder, Angels Deck (VC / business-angel club)🎙 Angela — host / interviewer; founder of House of Om, 22 years in the UAE🎙 Roman — director/editor, off-camera coaching cues (not in the cut)🎙 Roman's team / table voice — setup chatter (cut)
Igor Kaloshin is the CEO and co-founder of Angels Deck, a Dubai-based business-angel club six years old that has backed 180+ startups worth roughly $75-80M. In this candid home conversation with House of Om founder Angela, he explains how he hunts for 'Olympic-champion' founders, why 9 of 10 startups die, and the brutal math of venture returns. The spine of the episode is one honest failure: a deep-tech bet called GPU Audio where investors put in about $3M, the team built a product the whole music industry loved, and it earned just ~$100K because the market was too small — until they pivoted into automotive and turned the right market into a potential 50x. Igor breaks down the mistakes founders make (loving the product, not the customer's pain), real portfolio numbers (5 exits at 3-13x, hunting for 50x), why scaling can kill the soul of a business, and where he's putting money next (AI-native companies, seed stage). A practical, numbers-honest masterclass on building and betting in the UAE.
~121mraw source
~41mpremontage cut
14kept blocks
12chapters
14,332words
Story structure — stages of the video: where & what is discussed
1
HOOKThe $3M lesson🟢 keep
🎙 Igor Kaloshinsrc 1143 · 43:37–44:17▸ chapter
Cold-open pulled from the middle: Igor reveals a startup his club backed spent about $3M, earned only ~$100K, and what that taught him about choosing the wrong market. Hard number, instant stakes.
43:41 (1143)“They actually spent $3 million. They earned something like $100K. and you know it's not comparable. And this is a lesson that shows us that you might have a good product, but if you have chosen not the right path or not the right market, you might fail.”
2
INTROMeet Igor — 180 deals🟢 keep
🎙 Igor Kaloshinsrc 1143 · 4:50–5:54▸ chapter
Angela invites Igor to introduce himself; he frames himself as a venture capitalist and gives the headline numbers — Angels Deck, six years, 180 deals, ~$75-80M deployed, ~$300K per deal. Sets credibility and the promise of the video.
4:54 (1143)“I'm Igor and I'm a venture capitalist. We find the best and brightest early-stage startups, with the hope that at some point they will become unicorns and we will become a bit richer than now.”
5:26 (1143)“Six years ago we created a business angels club called Angels Deck, and since then we invested in like 180 deals worth about 75, probably $80 million.”
3
THE MODELSchool of investors🟡 trim
🎙 Igor Kaloshinsrc 1143 · 5:57–7:11
Igor explains what makes Angels Deck different from a classical fund: members make their own invest/no-invest decisions, learn fast by reviewing deal memos, and many go on to start their own clubs. 'Not just an investment vehicle, but a school of investors.'
7:14 (1143)“I would say we are not just an investment vehicle, but we are also the school of investors.”
4
THE BUILDWhat makes a unicorn🟢 keep
🎙 Igor Kaloshinsrc 1143 · 10:33–13:42▸ chapter
The core thesis: startups are built for enormous scale, not regular business. The Olympic-champion analogy — 5-9 years of extreme training, 2x growth a year compounding to 32x/100x, and why product is only ~5% while sales and scale are 95%. Strong team + scalable market.
11:47 (1143)“A lot of people work out, they like sport — that's similar to regular business. But if you want to grow up to Olympic champion, you need to train every day.”
12:44 (1143)“You need a product, but product is five percent of the business. 95% is sales of the product and scaling it. We expect growth in company capitalization of at least 2x a year. In five years it will be 32x. In seven years more than 100x.”
5
MISTAKESWhy 9 of 10 fail🟢 keep
🎙 Igor Kaloshinsrc 1143 · 14:17–16:02▸ chapter
Angela cites the 9-in-10 failure stat; Igor reframes it as the rule of the game, not 'doing it wrong' — it's a fight to be world champion. 90% fail or become zombies; one in ten returns the whole portfolio. The most common founder mistake: loving the product, forgetting the customer's pain.
15:22 (1143)“You are fighting to prove that you are able to be the champion. We understand that 90% of startups will fail or will become a zombie.”
16:12 (1143)“One out of 10 might be the star that will return the money invested in all 10 startups and give extra profit. And this is how it works.”
6
THE PITCHSell the pain, not the product🟡 trim
🎙 Igor Kaloshinsrc 1143 · 26:05–31:02▸ chapter
Igor's most actionable advice for founders: don't talk about the product, talk about the customer's pain and why they'll pay. Get into accelerators (HUB71, RSA Capital's Sandbox, DIFC). Investors fund the path to a big venture, not feature lists.
28:16 (1143)“There are a bunch of good accelerators here in UAE — specifically HUB71 in Abu Dhabi. RSA Capital has their own accelerator called Sandbox. There are a couple of accelerators in DIFC.”
30:23 (1143)“Sometimes I ask founders not to talk about their product at all. Let's start with the pain of the customers that you are solving. Try to explain the pain first.”
7
THE BETGPU Audio — the $3M story🟢 keep
🎙 Igor Kaloshinsrc 1143 · 37:27–43:37▸ chapter
The heart of the episode, told in full. Igor explains deep-tech, then the GPU Audio story: a startup that can parallel-process audio on a tiny GPU chip, building per-passenger 'sound capsules' inside a car so driver and back-seat child don't hear each other. Beautiful product, wrong market.
37:30 (1143)“It's not so important what you make, but what you make possible. We have a deep-tech startup called GPU Audio.”
39:23 (1143)“They can create a so-called sound capsule for each separate passenger. You're in the driver's seat using hands-free, and on the back seat your child is watching cartoons, and you don't hear each other.”
8
THE PIVOTWrong market, right pivot🟢 keep
🎙 Igor Kaloshinsrc 1143 · 42:20–46:00▸ chapter
How the same company recovered: it found the company three years ago, backed it into the pro-audio market with other investors (~$3M), built loved products but a tiny market — then pivoted to automotive where demand and tech finally meet. The lesson: good product + wrong market = failure.
42:38 (1143)“We invested together with others around three million dollars, and they successfully burned it. They build amazing products. Everybody in the music industry loves them. The problem is that the market is not so big.”
44:17 (1143)“When they ran out of money, they decided to move into the automotive segment. There is huge demand there for this multi-zone functionality — nobody can make it except our guys. This is where market demand and product advantage meet.”
9
SCALEScale can kill the soul🟡 trim
🎙 Igor Kaloshinsrc 1143 · 64:17–66:20▸ chapter
Angela asks how to scale House of Om; Igor uses the premium-restaurant analogy — you can't clone the chef, and scaling can cost you the soul. The honest counter-intuitive beat: bigger isn't always better, but unit economics can justify a tighter, premium model.
65:00 (1143)“Do you really need to be bigger? You can't clone the chef. So by scaling your business, you might lose the soul.”
10
NUMBERSHonest returns & exits🟢 keep
🎙 Igor Kaloshinsrc 1143 · 81:12–85:17▸ chapter
The honest-numbers beat viewers come for. Six years in, mid-cycle: lots of losses already, good traction from star companies, and so far 5 exits at 3-13x. What they're hunting for: 50x from one star company becoming a unicorn. Plus the trap — more expertise makes you more demanding, less opportunistic.
81:32 (1143)“So we went through those losses a lot already. We already had a couple of successful exits — five if I'm not mistaken. They're not huge, like three, five, maximum 13x on investments. What we are looking for is something like 50x at least from one of our star companies when they become unicorns.”
83:28 (1143)“The more experience you have in selecting startups, the fewer companies you can treat as potential unicorns. You become more and more demanding and less opportunistic.”
11
STARThe startup that changes lives🟢 keep
🎙 Igor Kaloshinsrc 1143 · 85:46–88:22
Igor's favorite portfolio story: Educate Online (now Fly Lane) lets K-12 schoolers study online in leading US/UK/Canada/Australia schools for ~$5K/year vs ~$100K offline — opening top-tier education to far more families. A clean, emotional 'this is how startups change the world' beat.
85:46 (1143)“We have one of our portfolio stars, a company previously called Educate Online, now Fly Lane. They allow schoolers from K-12 to be educated online in leading schools from the US, UK, Canada and Australia — about 20x cheaper than sending your child physically.”
87:39 (1143)“Offline it would cost approximately $100K a year; with Educate Online they get the same level of diploma for $5K a year. This is how startups change the world.”
12
NOWWhere the money goes next🟢 keep
🎙 Igor Kaloshinsrc 1143 · 90:00–92:12▸ chapter
Igor's investment focus going forward: it's late to back the giant LLM builders, but there's a blast of companies using AI to transform traditional business — that's what they hunt. They invest at seed (proof of concept, ~$50K/month revenue) and follow on through to exit.
90:00 (1143)“It's a bit late to invest in giant companies like OpenAI. But there is a huge blast of companies using those LLMs to dramatically change the businesses we see right now — and of course we are hunting for those.”
91:36 (1143)“We mostly invest in the seed stage where there's already proof of concept and some early revenue, like 50K a month, and we love to follow on the following rounds.”
13
RAPID FIREQuick-fire round🟢 keep
🎙 Igor Kaloshinsrc 1144 · 5:54–9:00▸ chapter
Continuation in clip 1144 — the rapid-fire round. Biggest founder mistake (not listening to customers); most impressive recent pitch (Reiner, armored-car healing in Brazil); the company he'd join (Pinscreen, the 'AI Pixar' changing film); and the single biggest red flag in a founder.
6:16 (1144)“What's the biggest mistake a founder has ever made? Not listening to the customers.”
8:25 (1144)“If you ask me what company I would love to be a co-founder of, I would say Pinscreen — because they are changing the film industry, and they will change it completely. It would be the AI Pixar of the future.”
14
ADVICEThe biggest red flag🟡 trim
🎙 Igor Kaloshinsrc 1144 · 12:44–23:52▸ chapter
The strongest single insight in 1144 and a clean closer/advice beat: the red flag that tells Igor a founder doesn't understand venture capital — 'we need money right now and after we build the product we won't need any more funds.' Building a unicorn needs many rounds; you can't grow like crazy on your own money.
12:44 (1144)“One of the biggest red flags is when I tell founders to come later, at the next stage, and they respond: but we need money exactly right now, and when we build the product we will not need any funds from you. This shows me they don't understand how venture capital works.”
13:22 (1144)“To build a unicorn you need many rounds of investment to fuel your exponential growth. It's almost impossible to grow like crazy having only your own money.”
23:46 (1144)“Let's leave something for humanity — some kind of islands without AI.”
YouTube chapters — EN, from the assembly timeline
0:00 The $3M lesson
0:37 Meet Igor — 180 deals
2:49 What makes a unicorn
5:50 Why 9 of 10 fail
7:31 Sell the pain, not the product
12:09 GPU Audio — the $3M story
17:52 Wrong market, right pivot
19:31 Scale can kill the soul
21:21 Honest returns & exits
27:29 Where the money goes next
29:38 Quick-fire round
32:26 The biggest red flag
What to drop — sections to cut / trim
src 1143 · 0:00–4:54Pre-interview setup chatter: Angela's England trip, sound-healing bowls, House of Om backstory, 'I've got questions on ChatGPT', icebreaker talk. Cut to cold-open; the real intro starts when Angela invites Igor to introduce himself.
src 1143 · 5:54–10:35 (partial)Tangents on coffee mornings logistics, House-of-Om psychotherapy/ice-bath banter and director cues ('mention Dubai') — keep only the school-of-investors core (block 3), trim the rest.
src 1143 · 19:05–25:57Coffee-morning logistics (Dubai vs Abu Dhabi vs RAK), invite-only event chat, repeated director coaching ('mention some incubators', 'people can watch from around the world') and cat drama (Tiger Lily). Off-topic for the cut.
src 1143 · 45:54–64:27Long House-of-Om consulting drift — questionnaire for housemates, construction/prop-tech platform idea, MBZUAI university tangent, courses/Coursera, loneliness mission. Personal to Angela's business, not the VC story. Salvage only the scale/soul analogy (block 9).
src 1143 · 1:32:44–1:36:49 (end)Wrap, then full re-take attempts of the intro after Roman realises he needs a hook/beginning — multiple false starts ('Welcome to House of Om, Igor'), door/noise coordination, 'let's make a pause'. All production chatter.
src 1144 · 0:00–5:52Repeated intro re-takes and director coaching (Speaker 3): 'introduce Igor first', 'you need bullet points', glasses/door/noise coordination, 'let me get my ChatGPT'. The real quick-fire interview begins at ~5:52.
src 1144 · interspersedDirector/team cut-ins between quick-fire questions (Speaker 3/4/5: 'Angela you didn't get this company', 'ask what is deep tech', repeat takes of red-flag answer). Keep only Igor's clean answers; cut the coordination.
Notes
Speaker map by content: Speaker 1 = Igor (guest VC); Speaker 2 = Angela (host, House of Om, 22 yrs UAE); Speaker 3 = Roman (off-camera director/editor giving live cues — never in the cut). Speakers 4-7 = team/table/cross-talk noise, all cut.
This whole shoot was coached live — Roman repeatedly interrupts ('I need a hook', 'mention Dubai', 'ask simple questions'). Every one of those interjections is CUT material; they're the editor's own notes spoken aloud.
Cold-open (block 1) deliberately pulls the $3M/$100K line from 43:41 to open on stakes, per channel person-first hook rule. Block 2 (the intro at 4:54) becomes the on-screen 0:00 reference after the hook, but block 1 is the first frame.
GPU Audio = the load-bearing arc (blocks 7+8). Igor garbles 'pivot' as 'children' in ASR a couple of times ('if you have children not the right path' = 'if you have CHOSEN not the right path') — fixed in quotes. Verify spelling 'GPU Audio', 'Reiner', 'Pinscreen', 'Educate Online / Fly Lane', 'HUB71', 'RSA Capital Sandbox', 'MBZUAI' against source before lower-thirds.
Honest numbers to overlay (block 10): 180+ deals · ~$75-80M deployed · ~$300K/deal · 90% fail · 5 exits at 3-13x · hunting 50x. These are the figures this audience searches for — give them screen text.
B-roll the business names: GPU Audio (pro-audio → automotive cabin), Educate Online/Fly Lane (online K-12), Pinscreen (AI film/VFX), Reiner (armored cars Brazil), HUB71/DIFC establishing shots. The setting (home, table, cats) is its own texture per channel DNA — keep light cat moments only if they land.
Pacing: clip 1143 is one long take with Igor talking in 60-90s monologues — block 4, 7, 8, 10 all need B-roll or text overlays inside them to break the talking-head and survive the 5/8-min retention dips.
Total kept runtime is rich; if a tighter cut is needed, blocks 3, 9 and the front of block 14 are the trim candidates (marked 'trim'). Blocks 1,2,4,5,7,8,10,12,13 are the spine — don't cut.
Project files · sources · UXP brief — what this was built from
The UXP brief is the Assembly JSON your Premiere UXP panel ingests — every segment carries block / use / is_chapter / color / speaker + full text, timecodes clip-local per source_file. The transcript adds per-word {w,s,e} timing for word-based editing.
YTUAE01 · Igor Kaloshin · channel «UAE People» (YTUAE) · premontage / video structure on yt.rya.ae · R.Y.A Media Lab FZE. How it was made: camera audio → word-level transcript (mlx whisper large-v3, per-word timecodes) + speaker diarization (pyannote) → assembly brief (story blocks, chapters, cuts). Word-level timing enables word-based editing in Premiere.