The Runway Ventures

Get the weekly newsletter to help you learn from startup mistakes and become a better founder.

Every Sunday, we analyse and share:

• 1 failed startup ☠️
• 2 costly mistakes made ⚠️
• 3 lessons learned for founders 🧠

90% of startups fail because founders made mistakes. Don't be one of them.

Subscribe for free here 👉🏻 www.therunway.ventures

Founder

beehiiv

beehiiv empowers you to create, monetize, and grow your newsletter with an all-in-one publishing suite.

(right team, great product, crazy execution)

Angel Investor

Fathom - AI Meeting Assistant

Fathom records, transcribes, highlights, and summarizes your meetings so you can focus on the conversation.

(proud user of Fathom)

X10 Media

X10 Media, a regional influencer agency established over 7 years ago before Influencer Marketing became part of the mainstay, holds a network of over 3,000 Influencers with a combined 30 million network reach in Singapore, Malaysia & Indonesia.

The agency has served over 1,000 brands including Pepsi, Singtel, Capitaland, LVMH, Warner Music, Owndays, Casetify among others.

With its proprietary Leadgram technology and data-driven insights, X10 Media empowers brands to achieve exceptional results through influencer campaigns.

X10 media currently owns 3 business units — Faves, X10 Ventures & TheSpokenTribe, that span regionally across 3 countries in Singapore, Malaysia & Indonesia.

🌏 Faves (micro-influencer marketing arm) runs a combined network of 10 million reach in Singapore, Malaysia & Indonesia.

🌏 X10 Ventures (technology arm) empowers brands with our proprietary technology.

• Leadgram was launched as a first-of-its-kind software in Asia to deepen social media campaign ROI & sales/traffic conversions for brands & companies.
• We retarget consumers that engages on influencers/brand content to create conversations leading to higher call to actions.

🌏 The Spoken Tribe (LinkedIn marketing arm) brings professional voices to endorse companies and brands through the word-of-mouth recommendations.

👉🏻 With a commitment to excellence and innovation, X10 Media continues to drive impactful influencer marketing solutions for brands. Visit www.x10.media to learn more.

CTO (Acquired by TSL Media Group)

Admond Lee

On a mission to make data science accessible to everyone through consulting & corporate training.

Data science consulting and analytics services for startups, SMEs, and marketing agencies.

✅ Previous projects include:

• AI & marketing consulting
• Web analytics & data visualisation
• Web scraping & marketing analytics
• Crypto price prediction model using deep learning with millions of sentiment data points


✅ Corporate & Institutional Training (Data Science / Analytics):

• SPH, DBS, MINDEF, Singtel, NTU, NUS, UTAR, Decoded


▶️ Learn more at www.admondlee.com

Consulting Data Scientist | Corporate Trainer

chatted.ai

Get AI insights from data in seconds and make informed decisions 10x faster.

1. Upload your data
2. Ask questions
3. Get actionable insights

That's it.

❌ No more formulas
❌ No more pivot table
❌ No more learning how to code
❌ No more waiting for your reports

✅ Just ask.

Sign up for free here 👉🏻 chatted.ai

... and start chatting today 🧠

YouTube

Making videos focusing on data science, programming, and tech stuff in general.

👉🏻 Link to channel: youtube.com/@admond1994

Content Creator

Staq

At Staq, we are building the universal API to help fintech companies securely access financial data from SMEs in a quick and easy way across Southeast Asia.

We are reimagining the technical infrastructure to drive financial inclusion for businesses in Southeast Asia.

Our mission is to empower you to build amazing financial products for businesses with more visibility and informed decisions. 

With Staq:
• You can easily and securely access financial data from businesses for SME lending, user identity verification, compliance and KYC etc. 
• Businesses can instantly share their data instead of sifting through tedious UX from multiple banks to extract bank statements. 

Headquartered in Singapore, Staq is backed by Entrepreneur First, the leading startup incubator in the world.

Co-Founder / CTO

Entrepreneur First

Entrepreneur First is the place where the ambitious come together to build globally important technology companies.

Founder in Residence (EFSG10)

Medium

Contributing Writer | Towards Data Science

Co-Founder & CTO @ Staq | Data Scientist

I’m a physicist turned data scientist, now a founder building The Runway Ventures — a weekly newsletter to help people learn from startup failures to become the top 1% founders.

I spent the first 3 years of my career helping companies solve business problems as a data scientist. Meantime, I was also a corporate trainer, speaker, and writer in data science. Then, I spent the next 3 years building startups. 1 failed, 1 acquired.

🧗🏻 This is how I got here:

• Got my degree in Physics from Nanyang Technological University.

• Started as a data scientist at SME (Titansoft), then MNC (Micron) before joining a startup (Hackwagon).

• Quit my job during COVID (wild timing), joined an incubator, found a co-founder, and built my 1st startup (Staq). Spent 1.5 years building Staq but we failed. Despite our mistakes, we learned a lot.

• Was very lost after my 1st startup’s failure. Spent the next 2 months doing self-reflection and exploring my curiosity.

• Started my 2nd startup building Leadgram at an influencer marketing agency (X10 Media). 8 months later, we got acquired by The Smart Local (TSL) Media Group.

***

After 3 years of building startups, I'm convinced that mistakes are the best teachers to become a better founder. For any startup, there are many ways to succeed, but only a few ways to die.

That’s why I started The Runway Ventures. 

Every Sunday, I analyse and share 1 failed startup story, 2 mistakes made, and 3 lessons learned with actionable tips to help you learn from startup failures.

Real life case studies, with key takeaways you can apply in your business.

No theories or BS.

99% of startups failed because founders made mistakes.

You’ll learn why and how startups failed to become the top 1% founder.

🤝🏻 Join us here (it’s free!): www.therunway.ventures

🔗 FB: @admond1994
🔗 Twitter: @admond1994
🔗 Instagram: @admond1994

Admond Lee Kin Lim

💍 She said YES!!!

After being together for more than 5 years, today, I proposed to her.

I still remember when I started my first startup. I was working as a food delivery rider part-time. I was broke af. I had nothing, literally nothing — except fire in my belly to make something people want.

It was the darkest period in my life. Yet, she supported me wholeheartedly throughout the time. She loved me unconditionally.

During that period, I told myself, “I want to marry her. I better don’t fxck this up.”

5 years later, here we are.

I’m sharing this with you because I really want to share the joy and happiness with you. 

You’ve been reading failed startup stories from me, and most startups failed because the founders made the wrong decisions.

Well, guess what?

💟 The most important decision of your life is to decide who you marry — your life partner. 

I’m glad I’ve found mine.

li_at

JSESSIONID

Helping you become the top 1% founder by learning from startup failures | 2x founder | 1 exit | Founder @ The Runway Ventures 🚀

$60M run rate. 5M+ downloads. #2 in Indonesia. Dead overnight.

Perfect metrics couldn't save Sorabel when they only had 6 months of runway.

Picture this: Indonesian husband-wife duo starts selling clothes from their garage in 2014. By 2019, they're sitting on 800 employees, 58% gross margins, and investors lining up for Series C.

But here's what should've terrified everyone:

They NEVER had more than 6 months runway. Even after raising $27M from Gobi Partners and Alpha JWC.

Every month was a race to raise more cash before the music stopped.

They even cracked the trust problem that kills most e-commerce in Southeast Asia. "Try First, Pay Later" — couriers wait 15 minutes while customers try on clothes at home. Only pay for what you keep.

Like a fitting room at your doorstep. Genius, right?

The numbers backed it up — $5M monthly revenue, 2.4x growth after rebrand, and they'd captured 9% of Indonesia's women's fashion e-commerce market.

CEO Jeffrey Yuwono was talking about becoming the Zara of Southeast Asia. Everything looked unstoppable.

March 2020: COVID hits. Investors can't fly to Jakarta. Term sheets pulled overnight. Fashion demand drops 50-70%. Their middle-class customers — the ones buying trendy clothes — all getting laid off.

July 2020: Game over. 800 jobs gone. Dreams dead.

Did COVID kill them, or were they already dying?

Tonight at 8pm, I'm publishing the full Sorabel autopsy. The 2 critical mistakes. The warning signs. Every lesson you need to know.

Subscribe now to get it when it drops → https://lnkd.in/gdDrhgtq

P.S. Innovation without unit economics is just expensive suicide.

$1.4B unicorn faked 4.8x revenue. $40B wiped in 3 days. CEO diverted funds to personal accounts.

Southeast Asia's fraud epidemic just hit critical mass — and 99% of founders are still blind to it.

I've written 90+ failed startup stories. But these 5 fraud cases shocked Southeast Asia:

eFishery (Indonesia's agritech unicorn):
• Claimed $752M revenue → Actually $157M
• Showed $16M profit → Actually $35M loss
• 6 years of dual-book accounting before whistleblower exposed it

LUNA/UST (Terraform Labs):
• $119 peak → $0 in 72 hours
• $40B market cap evaporated
• Millions lost life savings globally

Investree (P2P lending):
• CEO allegedly diverted company funds to personal account
• NPL hit 16% (reg limit: 5%)
• License revoked October 2024

The pattern is brutal: Fake metrics → Fake growth → Dead startup.

Singapore just dropped from #1 to #10 in global fraud resilience. Malaysia fell 52 places. Indonesia ranks dead last in Southeast Asia.

Your dashboards are lying to you. That 30% MoM growth? Could be bots. Those 10,000 signups? Might be fraud farms. Your growth might be fiction.

Nov 19-20, I'm at What the Fraud Summit (hosted by Sumsub) in Singapore. Because knowing these patterns before they hit you is worth more than your next funding round.

2 days learning defense beats 1 fraud scandal destroying everything.

Register here → https://lnkd.in/gN_TXqju
Use code ADMOND15 for 15% off.

P.S. If you think fraud only hits "other" startups, you're already a target.

Sumsub

3 Malaysian founders. 2 weeks. 4 all-nighters. 100,000+ users.

The LinkedIn Unwrapped story nobody's talking about.

Ashvin Praveen and his co-founders (Lizzie Tan, Muhammad Fathy Rashad) at Cleve.ai were behind on growth targets. Desperate. Running out of runway. They had this wild idea — build a "Year in Review" for LinkedIn creators.

Everyone thought they were crazy. Including themselves.

They shipped it anyway. In 14 days flat.

Week 1: 30,000 users (they expected 3,000)
Week 3: 100,000+ users
Countries reached: 100+

Even the CEO of PepsiCo Latin America posted their LinkedIn year-in-review. Alex Hormozi commented on their posts.

But here's what you don't know:

The early Cleve was "absolutely unusable." Users couldn't figure out the UX. 70% were on mobile but the product was built for desktop. Growth was dead.

So they did what most founders won't — burned the whole codebase and rebuilt from scratch.

Their 2 biggest fxckups:
1. Built a product that was "too clever" to use (tried to reinvent UX instead of following patterns people already know)
2. Spent hours in Mixpanel trying to "find the answer" in data instead of talking to users and trusting their gut

From 0 to 20,000 users grinding.
From 20,000 to 100,000+ in weeks.

All because they stopped patching problems and made the hard call.

Tonight 8pm: Why rebuilding from scratch saved them. The UX disasters that nearly killed them. How LinkedIn Unwrapped changed everything.

Subscribe now to get it when it drops → https://lnkd.in/gdDrhgtq

P.S. They rewrote their entire codebase on their one-year anniversary. Most startups die before making that call.

20,000 users → 100,000 users in 2 weeks.

Cleve.ai's founders pulled 4 all-nighters to ship LinkedIn Unwrapped. Even Alex Hormozi commented on their posts.

They almost didn't make it this far.

Ashvin Praveen admitted their early product was "absolutely unusable". Most people couldn't figure out how to use it. They tried to be too clever with design. Users dropped off instantly.

By late 2024, they'd crawled to 20k users but were missing growth targets. Then they discovered 70% of users were on mobile — and everything was broken.

The turning point: deleting everything and starting over.

Most founders patch problems. Cleve's team torched their codebase and rewrote it mobile-first. Scary? Yes. Necessary? Absolutely.

LinkedIn Unwrapped went viral because they'd finally built something that actually worked.

From university students running a personal branding agency to 100k+ users across 100 countries. All because they chose to rebuild instead of quit.

How they went from "absolutely unusable" to 100k users (and the 2 UX mistakes that nearly killed them):
https://lnkd.in/gzx2wQtP

P.S. Sometimes the best product decision is admitting your product sucks and starting over.

$53 million in customer funds. Vanished. Coinbase was about to drop $40M at a $400M valuation.

Then Zipmex collapsed overnight.

Marcus Lim had already exited OneFlare. Akalarp Yimwilai was Thailand's regulatory heavyweight — Georgetown master's, public admin doctorate, plugged into the Ministry of Finance.

Perfect split. Marcus raised money. Akalarp got licenses.

By August 2021, they'd pulled in $41M from Krungsri Finnovate (Bank of Ayudhya/MUFG), B Capital, Plan B Media. The "Coinbase of Southeast Asia" was crushing it.

Users loved the 10-14% yields on ZipUp+. Move your crypto from Thailand to Singapore entity. Zipmex lends it to Babel Finance and Celsius. You get fat returns.

Except nobody told users their money was actually sitting with leveraged crypto funds.

June 2022: Celsius freezes withdrawals.
Five days later: Babel halts redemptions.
Zipmex exposure: $48M with Babel, $5M with Celsius.

July 20: Zipmex freezes all withdrawals.

The rescue attempt was even worse. V Ventures promised $100M to buy 90% of Zipmex. They missed the first payment. Then the second. Then the third. Then they walked.

Court recovery plan: 5.1% back to creditors.
Straight liquidation: 1.6%.

February 2025: Akalarp gets 5 years in prison for fraud. Thai SEC reported that he moved customer funds offshore before updating terms & conditions.

The entire "regulated, licensed, safe" crypto exchange was just a funnel to risky lenders. When Terra blew up and took down 3AC, the dominoes fell fast.

Tonight 8pm: The full Zipmex autopsy drops. Every red flag. Every warning sign. The exact playbook of how "regulated" platforms blow up customer money.

Subscribe now → https://lnkd.in/gdDrhgtq

P.S. 14% yields in crypto? That money wasn't sitting in a vault. It was funding degens trading 10x leverage. You were the exit liquidity.

Last week, 35+ founders gathered to share their biggest fxckups. No BS. No sugarcoating. Just raw startup failures served with dinner and drinks.

We called it Startup Fxckups: Media Edition.

Willis Wee sold Tech in Asia for 8 figures. Joel Wong built Social Jet to 7 figures with 1,500+ creators. Both stood on stage and admitted: "We fxcked up. A lot."

Willis's confession hit hard:

"I ignored users for too long. Built features nobody wanted. Nearly killed the company."

His survival playbook:
• Ask users obsessively — stop assuming, start listening
• Watch your runway like a hawk — tough times always come
• When looking for best hires, ask HOW questions to know their thought process, review their case studies and do role play.
• Fire fast — 3 months max if they're not the right fit
• Test different ideas, remove what doesn't work, and double down on what works.

Joel doubled down on his biggest lesson.

His hiring revelation:
• The best hires with zero experience outperformed the "expert" every time
• Proactive beats qualified
• They fail fast, learn faster, grow fastest

50 founders. 2 hours. Zero pretense.

This is why we host these events. Not for the LinkedIn posts. Not for the networking cards. But because 90% of startups die from preventable mistakes.

And the only cure? Learning from founders who survived theirs.

Every founder left with at least one mistake they won't make now.

Massive thanks to Wong Wei Jer for co-hosting and making this happen. And to SUNRATE for hosting us and believing in honest founder conversations.

Want to join our next Startup Fxckups event? The only way to know it firsthand is through our newsletter → https://lnkd.in/gdDrhgtq

P.S. Best quote of the night: "If you're not embarrassed by your fxckups, they weren't big enough to learn from."

10 investors. 10 rejections. $10k left. 2 months to live.

I almost became a paper millionaire at 28. Instead, I burned my startup to the ground.

We had everything lined up for Staq. $100k pre-seed in the bank. Singapore's largest digital lenders ready to pilot. Term sheets being drafted for a $1M seed round.

I thought we were unstoppable.

Then reality hit.

The rejections came fast:
• 5 straight "NO"
• 3 investors pulled out: "This space is too tough"
• 2 wanted "more traction" (translation: you're not there yet)

We were burning $5k monthly. Bank balance: $10k.

Do the math. That's 2 months of runway.

Game over.

Here's the brutal truth nobody tells you about raising money:

Investors sense desperation. The moment you NEED their money, you've already lost. We were begging, not negotiating. They knew it. We knew it. Everyone knew it.

The 3 mistakes that killed us:

1. We built a solution looking for a problem. Not the other way around. No amount of VC money can fix that core problem.

2. I wanted to get rich more than I wanted to solve problems. Wrong motivation leads to wrong decisions, always.

3. We tried to raise money from a position of weakness. Never fundraise when you're desperate. Raise from a position of strength.

The shutdown was brutal. Telling customers we're done. Watching 2 years of work disappear.

But here's what failure taught me:

Building a startup isn't about the money. It's about caring so deeply about a problem that you'd solve it for free. Everything else — funding, growth, success — follows.

That's why I built The Runway Ventures. Every week, I dissect a startup failure so you don't have to make the same expensive mistakes.

My first startup cost me $100k to learn these lessons.

You can learn them for free.

Join 20,000+ founders getting the autopsy reports → https://lnkd.in/gdDrhgtq

P.S. The best time to raise money? When you don't need it. The worst time? When you're 2 months from death.

What's the most expensive lesson your startup taught you?

I've written ~100 failed startup stories.

My goal? Run out of them.

Sounds backwards, right? Most creators want endless content. I want to exhaust my supply.

Every Sunday, I dissect another collapse. Style Theory's $25M implosion. eFishery's $1.4B collapse. Bluelearn returning 70% to investors.

Each story means another founder learned what not to do.

When I run out of failures to write about, it means founders stopped making the same preventable mistakes. No more burning through SoftBank millions on broken unit economics. No more chasing vanity metrics while burning cash.

The graveyard is already full. 90% of startups die from self-inflicted wounds.

But if just 100 founders avoid one critical mistake from reading these autopsies, that's 100 companies that survive their first crisis. 100 teams that keep their jobs. Millions in investor capital preserved.

Running out of content means winning.

Until then, I'll keep writing. Every collapse. Every scandal. Every preventable death.

Because the best education costs someone else $25 million, not yours.

Join 20,000+ founders getting these stories before they make the mistakes → https://lnkd.in/gdDrhgtq

P.S. The goal is making this newsletter obsolete. Until then, every Sunday at 8pm.

$400M valuation. Coinbase circling. Thai bank backing.

Then 95% of customer funds vanished overnight.

Zipmex promised 14% returns. They delivered Southeast Asia's most catastrophic crypto collapse.

Thai users moved funds to "Z Wallet" for higher yields. They thought they were using a regulated platform. Reality? Their money crossed to Singapore, then straight to Babel Finance and Celsius.

$53 million. Gone.

The Thai co-founder just got 5 years in prison. He allegedly moved customer assets offshore before updating the terms & conditions.

I just published the complete Zipmex autopsy. The 2 fatal mistakes. The offshore deception. Why "regulated" meant nothing when $53M disappeared.

Read the full story → https://lnkd.in/gv8AbzRz

P.S. If someone promises you 14% guaranteed returns in crypto, ask yourself: where's that money actually coming from? Zipmex users found out too late.

Profitable cloud kitchen. Major hotel contracts. Big-name clients like Lazada and Maybank.

Still had to sell. Still burned out.

Mojos Kitchen's founders were waking up at 2am to prep breakfast boxes. By 6am, hundreds of meals were out the door. Nasi lemak, chicken curry, Western omelettes — all from one kitchen.

They'd survived COVID. Pivoted from physical kiosks to cloud delivery. Built up to 1,000+ app users. Finally hit profitability in early 2022.

Then sold the whole thing.

Not because the business failed. Because the founders were done.

Shrayas Saranathan told me something that stuck: "We wanted to prove the model first — turn profitable, then pitch. The only thing that happened was that we got burnt out first before we could go for funding."

The F&B grind is brutal. You're competing on razor-thin margins while managing:
• Multiple cuisines simultaneously
• 2am wake-ups for hotel contracts
• Staff turnover that never stops
• Equipment breaking at the worst times
• Compliance costs eating profits

They bootstrapped with over RM 1 million of personal money. Every ringgit reinvested. Zero networking. Zero investor meetings. Just heads down, grinding.

By the time they needed help, they had no one to call.

My parents run a hawker stall. I've watched them work 14-hour days for a decade. F&B isn't just hard — it's relentless.

Mojos Kitchen proved their model worked. They just couldn't survive long enough to scale it.

Full founder interview and autopsy here → https://lnkd.in/gR_Ui2_p

P.S. If you're running F&B and still standing, you have my respect 🫡

I tried Replit to build an MVP. I'm still processing what just happened.

20 minutes. Plain English. Working app. On my phone.

I've been skeptical of no-code tools for years. They promise the world, deliver toys. So when someone said "just try Replit", I rolled my eyes.

Then I had a problem that pissed me off.

We have 20,000 founders in our TRV community. When someone emails me asking for help, I spend hours trying to find the right founder with relevant experience. Manual search. Memory gymnastics. Usually give up.

Yesterday, I opened Replit.

Typed: "Build an app where founders can ask questions and get matched with the right founder based on their profiles, company stages, and experience."

I expected errors. Setup hell. "Please configure your API keys."

Instead? 20 minutes later, it was live.

The app:
• Analyzes 20,000 founder profiles automatically
• Matches questions to expertise in seconds
• Connects founders instantly

Zero code written. Zero.

I just described what I wanted. In plain English. Like texting a friend.

It connected to our Notion. No OAuth dance. No API documentation. Just logged in and it worked.

Deployment handled. Security built in.

The old way: Find developer. Negotiate price ($5K-10K). Write 20-page specs. Wait 6 weeks. Debug for another month. Still doesn't work right.

The new way: Get pissed. Type what you want. Ship it.

I'm still skeptical of most no-code tools.

But Replit? I don't even know what to call this.

Can it build the next Uber? No. Can it solve 80% of your operational headaches? Absolutely.

Every founder has that one operational nightmare that drives them insane. You know the one. The thing you've been "meaning to fix" for months.

Stop living with it.

Build your first app today → https://replit.com

I invested in beehiiv in 2024. Best decision I've made.

$32M annual revenue. 3 billion emails monthly. Creators earning $3M monthly on platform. All in 4 years.

Love the investor updates from Tyler Denk 🐝. Real numbers. Real progress. No bullshit.

I'm not just an investor — I'm a power user. The Runway Ventures runs on Beehiiv. Every newsletter. Every subscriber. Every dollar we make flows through their platform.

Tyler says they're building the operating system for the content economy.

Watch what he shipped last month:
• Digital products
• AI website builder
• Native podcasts
• Link-in-bio
• Full analytics suite

10 features. One day.

Most founders would call that their 2026 roadmap.

The playbook is simple:
• Watch creators juggle 7 different platforms
• Build everything native in one place
• Ship faster than anyone thinks possible

Half their team runs newsletters on Beehiiv. Tyler himself grew Big Desk Energy to 125,000 subscribers. They don't build for creators. They ARE creators.

TIME runs on Beehiiv. TechCrunch runs on Beehiiv. Arnold Schwarzenegger runs on Beehiiv.

While competitors debate features in product meetings, Beehiiv ships them before lunch.

The content economy needs its operating system. Someone's going to own it.

My money's on the founder who ships 10 features while others write one roadmap.

I love you beehiiv.

Willis Wee sold Tech in Asia for 8 figures. Now he's spilling his biggest media startup fxckups.

November 27. Our FINAL Startup Fxckups event in 2025. And we're going out with a bang.

Willis built Tech in Asia from a university project into Southeast Asia's most influential tech media platform.

Joel Wong? Scaled from dropshipping streetwear to running 250+ influencer campaigns. Generated 7-figure revenue. 

The media/agency space is brutal. One wrong hire, one bad client, one misread metric — you're dead.

These founders survived. Most don't.

Forget polished panels. This is founders telling you exactly how they fxcked up so you don't have to.

Free dinner. Free drinks. Unfiltered truth about building in media.

26 founders are joining. Only 4 spots left. When they're gone, they're gone.

This is it. Our last one. Don't miss it.

Register now → https://luma.com/0774rm7j

P.S. If you're building anything content-related or in media space, you need to hear these stories. The graveyard is full of media startups that looked unstoppable.

P.P.S Oh, and don't forget to use "TRV20" for 20% off

Make something people want.

Tech in Asia

Helping Brands Scale With A Full-Funnel Influencer Strategy | Ex E-Commerce Brand Builder

Just read ALL LinkedIn influencer's predictions for 2026, so you don't have to:

OpenAI is dead.

AGI is dead. 

Em-dash? Em-dead

Economy? dead (tomorrow)

Employees? Automated

9 to 5 - somehow also dead?

996? Kills you 

Remote work is dead.

The office is however also dead.

SEO...dead, again. 

AIO start-ups... dead upon arrival

AI Wrappers in general ... oof dead

Seat-based pricing? dead

Pricing in general. Dead - Free is the new MOAT

ARR. Dead

Sales? Replaced by AI SDRs (who are also dead)

The EU? Dead (consent collected) 

Germany? Dead even before the EU 

Stay positive 🫶

They achieved profitability. Landed major hotel contracts. Then sold the company.

Because the founders were done.

Mojos Kitchen just proved that even profitable startups can fail when founders burn out.

Picture this: You're finally profitable after 3 years of grinding. Your Kuala Lumpur cloud kitchen serves thousands of meals daily. Lazada, Maybank, and a major international hotel chain are your clients.

You should be celebrating. Instead, you're waking up at 2am to pack breakfast boxes.

Gaurav Joshi and Shrayas Saranathan put in RM 1 million of their own money. Not investor cash. Their savings.

COVID killed their 8 kiosks across KL overnight. Revenue went to $0. Meat rotting in refrigerators for months.

They pivoted. Built a cloud kitchen. Climbed back to profitability.

But profitability came with a price:
• 2am starts for 6am hotel deliveries
• Producing Malay nasi lemak, Chinese noodles, Indian curry simultaneously
• Managing B2C app orders, corporate boxes, hospital catering all at once
• No time for their kids anymore

The brutal truth about bootstrapping in F&B:

They waited to raise after proving the model. Smart, right?

Wrong. By the time they proved it, they were too exhausted to pitch.

"We wanted to prove the model first — turn profitable, then pitch. The only thing that happened was that we got burnt out first before we could go for funding."

That's what Shrayas told me when I interviewed him.

Two critical mistakes killed them:
1. Zero networking while building (no investor relationships when they needed them)
2. Bootstrapping a capex-heavy business too long (RM 1 million hole they never climbed out of)

They had 4 interested buyers. 3 ghosted. 1 lowball offer they took just to escape.

Tonight at 8pm, I'm publishing my full interview with Shrayas and the Mojos Kitchen story. The hidden cost of bootstrapping RM 1 million in Malaysia's F&B scene that nobody talks about.

The business was winning. The founders were dying.

Get the complete breakdown when it drops → https://lnkd.in/gdDrhgtq

P.S. If you're bootstrapping in F&B, you already have my respect. But please, build your network before you need it.

74% of enterprise deals go to whoever responds first. Not the best product. Not the lowest price. The fastest.

I just ran a B2B enterprise sales masterclass with Ned Phillips for my private founder community. 10 tactical plays that actually close deals:

1. 𝗦𝘁𝗼𝗽 𝘀𝗲𝗹𝗹𝗶𝗻𝗴 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝘀
Spend 80% of calls listening. Only mention your product in the last 10 minutes — after they ask.

2. 𝗩𝗼𝗹𝘂𝗺𝗲 × 𝗩𝗲𝗹𝗼𝗰𝗶𝘁𝘆 = 𝗥𝗲𝘃𝗲𝗻𝘂𝗲
Target: 3 sales meetings/day. That's 30 warm outreaches + 100 cold actions daily. Manual, personal, thoughtful. Not spam.

3. 𝗟𝗶𝗻𝗸𝗲𝗱𝗜𝗻 𝗳𝗶𝗻𝗱𝘀 𝗱𝗲𝗮𝗹𝘀, 𝗲𝗺𝗮𝗶𝗹 𝗰𝗹𝗼𝘀𝗲𝘀 𝘁𝗵𝗲𝗺
Post daily about problems you solve (not your product). Comment on prospects' posts. Follow up within 5 minutes of any inbound lead.

4. 𝗨𝘀𝗲 𝗮 𝗖𝗥𝗠 𝗼𝗿 𝗱𝗶𝗲 𝘁𝗿𝘆𝗶𝗻𝗴
HubSpot. Every deal needs a next action assigned. No next action = dead deal.

5. 𝗞𝗶𝗹𝗹 "𝗺𝗮𝘆𝗯𝗲" 𝗳𝗮𝘀𝘁
"No" gives you clarity. "Maybe" wastes months. Walk away if their pain isn't painful enough.

6. 𝗬𝗼𝘂𝗿 𝗰𝗵𝗮𝗺𝗽𝗶𝗼𝗻 𝗰𝗮𝗻'𝘁 𝘀𝗲𝗹𝗹 𝗶𝗻𝘁𝗲𝗿𝗻𝗮𝗹𝗹𝘆
Write their email to their boss. Draft their internal deck. Remove 100% of friction. They just forward and click send.

7. 𝗥𝗲𝗰𝗼𝗿𝗱 𝗲𝘃𝗲𝗿𝘆 𝗽𝗶𝘁𝗰𝗵
Watch yourself for 10 minutes daily. Track: % time talking, questions asked, objections fumbled.

8. 𝗗𝗮𝗶𝗹𝘆 𝗻𝗼𝗻-𝗻𝗲𝗴𝗼𝘁𝗶𝗮𝗯𝗹𝗲𝘀:
• Find 1 new lead
• Engage 1 warm prospect 
• Follow up 1 existing deal
• Post once on LinkedIn
• 10 total sales actions minimum

9. 𝗧𝗵𝗼𝘂𝗴𝗵𝘁𝗳𝘂𝗹 𝗯𝗲𝗮𝘁𝘀 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗲𝗱
Ned's example: Someone sent him 5 curated running videos → instant meeting. Research manually. Send something personal.

10. 𝗦𝗽𝗲𝗲𝗱 𝘄𝗶𝗻𝘀 74% 𝗼𝗳 𝗱𝗲𝗮𝗹𝘀
Submit RFPs early. Respond same-day to emails. First to follow up usually wins.

The brutal truth? Most founders don't have enough pipeline. They're doing 10 outreaches when they need 100.

Enterprise buyers buy trust, speed, and likability. Not PowerPoints.

My private founder community gets you:
• Private WhatsApp community of vetted founders
• Quarterly private dinners with our members
• Concierge intros + 1-1 founder matching
• Monthly live masterclasses with top founders

Ready to level up? Apply to join:
https://tally.so/r/gDDeyN

P.S. Massive thanks to Ned Phillips for dropping these bombs. You're the GOAT in B2B enterprise sales 🐐

Willis Wee sold Tech in Asia for 8 figures. Joel Wong scaled to 7 figures, then built 250+ influencer campaigns. And they're about to spill their biggest media startup fxckups.

These aren't just any founders. They're the ones who actually survived the media bloodbath.

27 November. SUNRATE Singapore. We're hosting Startup Fxckups: Media Edition — where successful founders share the disasters that almost killed them.

This isn't your polished TED talk. No scripted success stories. Just raw, unfiltered truth about what it really takes to survive in media.

Willis built Tech in Asia from a university project into Southeast Asia's most influential tech media platform. Then sold it to SPH Media for an 8-figure exit.

Joel dropped out to build a streetwear brand. Scaled to 7 figures. Then started Social Jet Singapore - Influencer Marketing Agency — now running campaigns for the region's biggest brands with 1,500+ creators.

The media/agency space is brutal if you get it wrong. One bad pivot, one wrong hire, one missed trend — game over.

And right now, with AI eating content jobs and ad costs through the roof, the margin for error is zero.

That's why we're bringing them together. To share the expensive mistakes they made so you don't have to.

What you'll get:
• Their biggest fxckups + learnings
• The exact pivots that saved their companies
• The hiring disasters that almost ended everything
• How they scaled from $0 to 7-figure revenue (and finally got acquired)

Plus open mic session — share your own fxckups and learn from others in the trenches.

Dinner and drinks included. Capped at 50 people — serious operators only.

Use code TRV20 for 20% off → https://luma.com/0774rm7j

P.S. The last Startup Fxckups sold out in 3 days. 50 seats only. Once it's gone, it's gone.

SUNRATE

Social Jet Singapore - Influencer Marketing Agency

Which T-shirt would you wear? Design A or Design B.

I need your help.

After 2 years of building The Runway Ventures, we're finally making merch for our community. The designs are ready. But I'm stuck.

Both tested well internally. Both cost the same to produce. Both represent what we stand for — learning from mistakes to become better founders.

But I can only pick ONE.

You've been with me through ~100 startup failures. You get to decide what we wear.

Here's the deal:

Comment "A" or "B" below. Most votes wins. Takes 2 seconds. Seriously.

✌🏻

$15.3 million in fines. 163 migrant workers exploited. The biggest worker penalty in Australian history.

Sushi Bay went from a single Sydney outlet to 16 locations across 3 states. National brand. Packed restaurants.

Then the Federal Court found what was really happening.

$653,129 in unpaid wages. Falsified records. Workers forced to pay back part of their wages.

Fair Work Ombudsman Anna Booth:

"If you deliberately underpay migrant workers and try to cover it up with false or misleading records you will be found out and will pay a heavy price."

The price? Record-breaking penalties. $21 million in debt. Every outlet liquidated except one.

The full story reveals how cutting labor costs became a death sentence.

Read the Sushi Bay collapse → https://lnkd.in/g7V-t6F6

Every week I dissect startup failures so you don't repeat them.

No BS AI/ML Content | ML Engineer with a Plot Twist 🥷30M+ Views 📝

Done-for you Audience Building & Content Writing for AI/ML Founders that frees up 100% of your time, so you can focus entirely on building your company  
  
Is your work too important to ignore? 💌 https://tally.so/r/wk69VJ  (or shoot me a DM) 👉
  
It's 2025 and we're at the peak of the attention economy  
  
Where Authenticity is king, visibility is currency and people wanna buy from people

Not faceless dehumanized brands  
  
So, they told you LinkedIn was "free marketing" you slapped some posts and called it strategy  
  
You've poured hours crafting the 'perfect' post, only to watch it sink without a trace..  
  
Then it dawned on you, your technical brilliance isn't enough to draw a crowd  
  
Hell, maybe you've even tried mimicking influencers, scrambling for any traction..  
  
Your post got 500 f*cking views... from your friends...  
  
Being ignored is soul-crushing  
  
It's not your fault tough  
  
When you started your company, you didn't sign-up for a degree in an Audience Building, Copywriting and LinkedIn strategy 

And here you are, sacrificing yet another late night for your vision  

"Am I just sabotaging my efforts?"  
  
"Is my life-work meaningless?"  
  
Getting traction feels impossible  
  
You hoped hiring some marketer would save your weekends but..  
  
Tech is a code they can't crack  
  
They can only churn out the lame, soul-sucking crap we all ignore on sight  
  
With armies of founders flooding feeds, your voice dissolves into meaningless white noise  
  
But in the process, you've wasted more of your precious time explaining 101 tech concept and rewriting low-quality drafts 

Done being ignored?  
  
Let's make the world listen 
  
I can help you with:  

✒️ No BS AI/ML Content  
⭐ LinkedIn Profile Revamp
🎯 Website Copy Optimization  

I've done it before:

↳ For founders backed by YC, a16z, Sequoia, and Techstars
↳ For bootstrappers in LLMs, ML, Vector DBs, Data Analytics, and Engineering
↳ For AI and Machine Learning leaders at Google, Microsoft, Airbnb, and YouTube

If they could rise above the noise, so can you

Yeah, even if you're an introvert with a "boring" niche

Still with me?  
  
Amazing—You're amazing!  
  
Cheers,  
Paolo

My booking link: 💌 https://tally.so/r/wLW0oz (or shoot me a DM) 👉

Paolo Perrone

NVIDIA just made their best AI models free.

99% of developers have no idea.

While you burn cash on OpenAI credits...
NVIDIA open-sourced Nemotron.
Their entire agentic AI family 🤯

Free Models That Actually Ship:
• Nano: Runs on your laptop
• Super: Single GPU beast
• Ultra: Multi-GPU monster

What They Do:
→ Graduate-level reasoning
→ Math without hallucinations
→ Code that compiles
→ Visual reasoning that works
→ Tool calling. No loops.

The genius part:

They published EVERYTHING.
Models. Datasets. Techniques.
Actually open source.

Game Changers:
• 6x faster on new hardware
• No API limits
• Runs on a Raspberry Pi
• Commercialize it. No strings.

The uncomfortable truth:

You're waiting for GPT-5.
Smart teams ship with Nemotron.

Who's using it:
- Accenture
- CrowdStrike  
- ServiceNow
- Deloitte

The kicker:

Download from Hugging Face. Now.
Deploy tomorrow.
Zero license fees.

Forever.

They built enterprise NIM too.
(That costs money. Models don't.)

Still paying for API credits? 

♻️ Repost before OpenAI charges per thought

No BS AI/ML Content | ML Engineer with a Plot Twist 🥷50M+ Views 📝

I just finished Google's biggest AI Agents course ever.

250+ pages of production insights. 100% free.

Here's everything they released:

DAY 1️⃣ INTRODUCTION TO AI AGENTS (50 pages)

Most agent demos fail in production.
This covers the 5 levels of AI agents, from basic reasoning to self-evolving systems, plus security and architecture patterns that scale.
🔗 https://lnkd.in/gN9JaXgR

DAY 2️⃣ AGENT TOOLS AND MCP (50 pages)

MCP servers can add new tools without your approval.
Your poetry bot suddenly gains purchasing power overnight.
This breaks down tool design, the Model Context Protocol, and hidden security risks.
🔗 https://lnkd.in/gRWDSjBr

DAY 3️⃣ CONTEXT ENGINEERING AND MEMORY (70 pages)

Agents with zero long-term memory treat every session like meeting you for the first time.
Learn how to build memory systems that actively curate context, not just save conversations.
🔗 https://lnkd.in/gbgWJ4kH

DAY 4️⃣ AGENT QUALITY (50 pages)

Traditional QA doesn't work for agents. You can't unit test judgment.
This covers the 4 pillars of agent quality (effectiveness, efficiency, robustness, safety) and how to evaluate reasoning.
🔗 https://lnkd.in/g6DHcZDx

DAY 5️⃣ PROTOTYPE TO PRODUCTION (40 pages)

Building an agent takes 2 weeks.
Trusting it in production takes 6 months.
Learn evaluation gates, circuit breakers, and evolution loops that close the gap.
🔗 https://lnkd.in/gYTakhAK

Huge props to Antonio Gulli and Shubham Saboo for making this happen, and make it 100% free

250+ pages. Zero cost. All production-focused.

♻️ Repost to save a fellow AI engineer 6 months of trial and error

Sebastian Raschka, PhD just exposed the biggest secret in AI.

After dissecting 9 flagship LLMs, he found something nobody wants to admit:

We haven't invented a new architecture in 7 years.

The evidence is damning:
• DeepSeek V3: 671B parameters, still a transformer
• Kimi K2: 1 TRILLION parameters, still a transformer
• GPT-OSS: OpenAI's latest, still a transformer
• Llama 4: Meta's flagship, still a transformer

Every "breakthrough" model?
Same skeleton. Different makeup.

Here's what's actually happening:

DeepSeek V3/R1: Compressed attention (MLA) + expert routing (MoE)
Gemma 3: Sliding windows to fake longer context
SmolLM3: Removed position embeddings entirely (!)
Kimi K2: Just made DeepSeek bigger

The brutal reality:

• No new architectures since "Attention is All You Need" (2017)
• Every innovation = efficiency hack
• We're polishing, not inventing

671 billion parameters.
1 trillion parameters.
Same. Damn. Architecture.

While everyone's chasing parameter counts, Sebastian mapped the actual mutations:
• MoE: Use 37B of your 671B parameters
• Sliding attention: Pretend you have infinite context
• MLA: Compress your KV cache and pray

This isn't innovation.
It's optimization theater.

The most honest technical analysis you'll read this year.

Read it. Then ask yourself:
If we can't innovate past transformers, what happens when we hit the parameter wall?

♻️ Repost if you appreciate actual technical depth over hype

AI agent bootcamps charge $2,000 for this. You're getting it for free.

1️⃣ Design (6+ hours)
→ Langflow docs: docs.langflow.org
→ Travel planning agent template: https://lnkd.in/gmbWX5Mg
→ Agentic design patterns by Andrew Ng: 

2️⃣ Build (22+ hours)
→ LangGraph course: https://lnkd.in/g7raxjpC
→ LangSmith (modules 1-3): smith.langchain.com
→ Evaluation: https://lnkd.in/ghE-8EwA
→ Prompt engineering: 4 Methods of Prompt Engineering

3️⃣ Deploy (35+ hours)
→ LangChain Academy (modules 1-6): academy.langchain.com
→ MLOps specialization: https://lnkd.in/gmeb53UP
→ System design fundamentals: YouTube playlist
→ Distilabel for synthetic data: https://lnkd.in/gd_jJaJE

63 hours. $0.
The $2,000 bootcamp teaches the same stack with better marketing.

💾 Save this. It's the curriculum.
♻️ Repost to save someone $2,000.

Wanna build AI agents? Stop watching random tutorials.

Here's the actual learning path:

1️⃣ Design (6+ hours)
→ Langflow intro: langflow.org
→ Multi-agent workflow patterns
→ Build: Travel planning multi-agent system

2️⃣ Build (22+ hours)
→ LangGraph course: https://lnkd.in/g7raxjpC
→ LangSmith for tracing: smith.langchain.com
→ Evaluation metrics and prompt engineering

3️⃣ Deploy (35+ hours)
→ LangChain Academy: academy.langchain.com
→ MLOps fundamentals
→ LangGraph Cloud deployment

Total: 63 hours. If you follow the order.
6 months if you don't.

💾 Save this.  Start at Phase 1.
♻️ Repost if you've ever jumped straight to deployment and regretted it.

90,000 AI papers published on Arxiv.

Ilya Sutskever says you only need 27 to understand 90% of what matters.

I just went through his list. Here's what shocked me:

What's NOT on the list:
❌ No RAG papers (everyone's obsessed with retrieval)
❌ No prompt engineering papers (despite the hype)
❌ No LangChain or agent frameworks
❌ Nothing about fine-tuning optimization
❌ Zero papers on "emergent abilities"

What IS on the list:
✅ "Attention is All You Need" (obviously)
✅ "The Unreasonable Effectiveness of RNNs" (2015, still matters)
✅ "Scaling Laws for Neural Language Models" (this is the one)
✅ "Kolmogorov Complexity and Algorithmic Randomness" (wait, what?)

The pattern I noticed:

1️⃣ Ilya's not telling you to read papers about using AI.
He's telling you to read papers about how intelligence works.
Three categories dominate his list:
Fundamentals (the architecture breakthroughs)
Attention mechanisms
Residual connections
LSTMs and sequence modeling

2️⃣ Scaling (why bigger actually works)
Scaling laws
Pipeline parallelism
Training dynamics

3️⃣ Compression (the theory underneath)
Minimum description length
Kolmogorov complexity
Information theory

Here's the controversial take:
If you're only reading papers from 2023-2024, you're studying tactics.

Ilya's list is strategy:

"Attention is All You Need" is from 2017.
"Understanding LSTM Networks" is from 2015.
"Deep Residual Learning" is from 2015.
The foundations haven't changed.
We're just scaling them.

The most surprising paper on the list:

"The First Law of Complexodynamics"
It's not even about neural networks.
It's about how complex systems evolve.

This is what separates engineers who use AI from engineers who understand AI.

If you only read 5:

1) Attention is All You Need (transformers)
2) Scaling Laws for Neural Language Models (why GPT works)
3) Understanding LSTM Networks (sequence modeling)
4) Deep Residual Learning (why deep networks train)
5) Kolmogorov Complexity (the theory of intelligence)

♻️ Repost to save someone from FOMO-reading 90,000 papers

99% of AI engineers can't write a GPU kernel. 

This free CUDA course on GitHub changes that.

Zero fluff, pure metal GPU programming.

What you'll build:
• GPU kernels from scratch
• Production-ready Triton code
• Optimized matrix multiplication 
   (what makes GPT-4 fast)
• Direct cuBLAS/cuDNN integration
• Custom PyTorch CUDA extensions

The uncomfortable truth: 
Every 10x performance gain in AI comes from someone who understood CUDA.
Time to join them 👇

♻️ Repost to save someone from being a framework-only developer

🚨 Update: 800+ comments. My DMs exploded.
Can't reply to everyone. 
Grab it here: https://lnkd.in/eMQn_X-6
-------------------------------------------
Hamel's AI Evals toolkit disappears in 24 hours.

This isn't publicly available.

4 resources from the guy who:
→ Trained 2,000+ engineers on evals
→ Consulted for OpenAI, Google, Meta
→ Delivers high-value engagements for top-tier companies (ask dbt Labs why)

What you're getting:

📇 12 Evals Flashcards — the visual system for memorizing what matters
📘 The Complete Evals Guide — 144 pages of everything you need
📚 The RAG Evaluation Bible — retrieval metrics that actually work
📧 30-Day Email Course — builds evals that predict user satisfaction (free)

You won't find this on Google, GitHub, or HuggingFace.

It's in your DMs. If you act fast.

Last time I did this, 4,173 people commented. I can't DM that many again.

Comment "EVALS" + Repost ♻️ and I'll send all 4

24 hours. Then it's gone.

There's a special hell in AI/ML engineering.

It hits around Year 2.
I call it the "valley of almost useful."

You're past "Hello World" tutorials.
Not good enough for real jobs.
Stuck building demos nobody uses.

The daily torture:

Morning: "Today I'll implement that paper"
Noon: Your embeddings are somehow 7-dimensional
Evening: Stack Overflow has zero answers
Night: "Maybe I should go back to web dev"

The worst part?

You know just enough to be dangerous.
Just not enough to be valuable.

🔥 The 3 Levels of "Almost Useful"

Level 1: The Overengineering Phase
Every problem needs a neural network.
CSV parsing? Better use GPT-4.
Your hello world has 17 dependencies.

Level 2: The Imposter Syndrome Pit
Everyone else seems to ship AI products daily.
Your Twitter feed is pure pain.
"Just deployed my 5th AI startup this week!"
Meanwhile, you can't get embeddings to work.

Level 3: The Clarity Moment
You realize 90% of "AI Engineers" are just prompt engineers.
Most production AI is boring Python scripts.
The magic isn't in the model - it's in the problem.

Here's what got me through the valley:

1️⃣ Ship broken things
My first production model had 67% accuracy.
Still saved the company $200K/year.
Perfect is the enemy of deployed.

2️⃣ Join a startup, not FAANG
Startups let you touch everything.
Break things. Learn fast. Ship faster.
Big tech will teach you one tiny slice.

3️⃣ Build boring solutions
Stop chasing cutting-edge.
Most businesses need basic classification.
Not your 12-layer transformer architecture.

4️⃣ Document everything
Future you will thank present you.
Especially when debugging at 3 AM.
"Why did I set embedding_dim=7?"

5️⃣ Find your valley friends
The loneliest part is thinking you're alone.
You're not. We all went through this.
Some of us just hide it better.

The uncomfortable truth:

The valley of almost useful is necessary.
It's where you learn that AI isn't magic.
It's engineering.

And engineering is mostly debugging.
And crying.
But mostly debugging.

Still in the valley?
Good. That's where real engineers are made.

Not in bootcamps.
Not in courses.
In the trenches, shipping broken code that somehow works.

Welcome to the club.

P.S. That senior engineer who seems to know everything? They googled it 5 minutes before the meeting. We're all just better at pretending now.

♻️ Repost to help someone realize they're not alone in the valley

I pushed every agent framework until it broke. Most broke fast.

CrewAI vs LangGraph vs Agno vs PydanticAI vs Atomic Agents...
Plus SmolAgents, Mastra, and AutoGen.

The brutal truth:

Most devs hate the popular ones.

CrewAI and AutoGen? "Tricky to debug"
LangGraph? "Overly complicated abstractions"
SmolAgents' CodeAgent? "Cool idea, doesn't always work"

What actually matters:

🎯 Control vs Agency
Some frameworks let LLMs run wild (AutoGen)
Others keep them on a tight leash (Atomic Agents)
Guess which ones actually ship to production?

🛠️ Abstraction Level
High abstraction = Quick start, debugging nightmare
Low abstraction = More code, but you know what broke
Pick your poison.

💀 The Learning Curve
LangGraph: Steep as hell, but scales
CrewAI: Easy start, hits a wall
PydanticAI: Write everything yourself

The uncomfortable truth:
"AI agent frameworks are some of the worst forms of abstraction"

But here's what actually works:
• Agno for developer experience
• LangGraph for multi-agent at scale
• PydanticAI when you need control
• CrewAI if you're just prototyping

Half these frameworks won't exist in 2 years.
The winners? Whoever makes debugging possible.

Saved you 2 weeks of testing. You're welcome.
♻️ Repost if you've been burned by agent framework promises

a16z just published their AI coding report.

The math blew my mind:

AI coding costs $10k/year per developer.
But creates $100k+ in value.
That's $3 trillion in new GDP.

The entire economy of France. From a few Silicon Valley startups.

Here's what nobody's talking about:

Software development just shifted from pure people cost to people + compute cost.

Your junior developer in Bangalore?
Now competing with Claude at $10k/year.

The uncomfortable truth:

We're watching the end of cost-based outsourcing.

Not because AI replaces developers.
But because location doesn't matter when compute costs dominate.

I tested the numbers myself:

• 100k context window filled
• Claude Opus 4.1 reasoning mode
• 10k output tokens
• 3 queries/hour, 7 hours/day

= $10,500 annually

That's more than many offshore developers.

But here's the kicker:

The companies winning aren't cutting developers.
They're hiring MORE.

Why?

Because when productivity doubles, the math changes:
• 1 developer + AI = 2x output
• Cost per feature drops 50%
• Suddenly every project has positive ROI

The report shows clear winners already:

Cursor: $500M ARR in 15 months
Windsurf: $2.4B Google acquisition
Claude Code: Started a war with its own distribution channel

The pattern is obvious:

Lower barriers → More projects → More complexity → More developers needed

AWS didn't kill sysadmins.
WordPress didn't kill web developers.
AI won't kill software engineers.

It just changes what they do.

University CS programs?
Already obsolete.

Teaching algorithms when AI writes them faster?
Teaching syntax when AI handles implementation?
Teaching debugging when AI prevents most bugs?

That's like teaching calligraphy in the age of keyboards.

The real skill now:

Knowing when the AI is hallucinating.
Dragging it out of architectural holes.
Translating business needs into prompts.

The $3 trillion opportunity isn't in replacing developers.

It's in making every developer 10x more valuable.

While universities update their curriculums,
smart developers are already adapting.

Are you?

♻️ Repost if you think CS education needs a complete overhaul

I just watched Sebastian Raschka build the entire attention mechanism from scratch in 2 hours.

No abstractions. No "just use torch.nn.MultiHeadAttention."

Line-by-line Python until it clicks.

This is 's breakdown of how LLMs actually work.

Step 1️⃣ Basic self-attention (no weights) 
Shows you the raw math: how a model decides which words matter.
You see why "bank" in "river bank" gets different attention than "bank" in "savings bank."

Step 2️⃣ Add learnable parameters 
Now the model can learn what to pay attention to.
This is where it stops being a simple lookup and becomes intelligence.

Step 3️⃣ Causal masking 
The trick that makes GPT work: tokens can only attend to previous tokens.
Without this, the model cheats during training.

Step 4️⃣ Multi-head attention 
The breakthrough: run 8 attention mechanisms in parallel.
One head learns syntax. Another learns semantics. Another learns long-range dependencies.
This is why transformers work.

👉 What makes this different:

Most tutorials either:
Show you the math (incomprehensible)
Show you the PyTorch API (you learn nothing)

Sebastian does both.

You see the math implemented in code, step by step.

By the end, you understand why attention_weights = softmax(Q @ K.T / sqrt(d_k)) isn't magic.

It's just weighted averaging with learnable weights.

💡The moment it clicked for me:

Attention isn't complicated.
It's just "which words should influence this word's representation?"
Then you do that for every word.
Then you stack it 96 times (GPT-4).

That's it.

Traditional RAG: retrieve → generate → pray

Agentic RAG: retrieve → validate → retry → succeed

One is a pipeline. The other is a feedback loop.

I just built this Agentic RAG system with LangGraph + Milvus

It knows when it's wrong, and fixes itself

Here’s how:

Router: "Can I answer without retrieval?" 
✅ Yes: Skip to generation 
❌ No: Retrieve from Milvus

Grader: "Are these chunks actually relevant?" 
✅ Good: Generate answer 
❌ Bad: Rewrite query, loop back

Validator: "Does this answer match the sources?" 
 ✅ Yes: Return to user 
❌ No: Retry with different approach

It's a graph with feedback loops, not a one-shot pipeline.

The difference:
Traditional RAG commits to one path. No second chances.
Agentic RAG loops back, validates, catches mistakes, and retries until it's right.

You get a system that debugs itself before you see the output.

Why this stack:
LangGraph = control flow primitives (routing, looping, branching) 
Milvus, created by Zilliz = production-grade vector search at scale
Together: RAG that thinks instead of guesses.

Full implementation 👇

This is the RAG architecture pattern that will become standard.
♻️ Repost to save a fellow engineer from another week of RAG debugging

BFCM isn't won by the brand with the best products.

It's won by the brand that sees what's breaking first.

Most operators spend BFCM like this: 

9 AM: Campaign's crushing it, scale to $10K/day
11 AM: ROAS drops 40%, nobody knows why
2 PM: Still diagnosing
5 PM: $6K burned on a broken campaign

The strategy wasn't wrong. 

You just couldn't see the problem fast enough. 

Diagnosing "why is ROAS dropping?" right now: 

→ Open Meta Ads Manager
→ Check Shopify analytics
→ Pull GA4
→ Cross-reference Klaviyo
→ Export to spreadsheet
→ Try to connect the dots

20 minutes later you've burned $3K.

Triple Whale + Moby give you real-time answers from your complete data stack. 

👉 https://lnkd.in/gtgB8GJw

Moby has real-time access to all of it. 

Ask Moby: "Why did ROAS drop?"

Get: "iOS campaign CTR crashed 67% after creative refresh. Top SKU went out of stock an hour ago. Traffic's converting to lower-margin backup products."

Not a guess. The actual answer.

One brand caught this: 

The campaign was showing 187% ROAS. 
Moby a flagged 41% return rate. 
Actual ROAS after was returns: 73%. 

They were about to scale into a loss.

During BFCM, 20 minutes to diagnose = thousands in wasted spend. 

The brands that win can see problems in seconds, not hours. 

And yes, I am a Triple Whale partner.

👉 https://lnkd.in/gtgB8GJw

♻️ Repost if you've burned budget figuring out what went wrong

taplio

The Battle

Admond Lee Kin Lim

Paolo Perrone

Followers

Posts

Engagements

Likes

Performance ScoreAdmond Lee Kin Lim's LinkedIn Performance

Performance ScorePaolo Perrone's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Admond Lee Kin Lim

Paolo Perrone

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Admond Lee Kin Lim's LinkedIn Performance

Performance ScorePaolo Perrone's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreAdmond Lee Kin Lim's LinkedIn Performance