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

S$30. Age 16. Zero experience.

That's how Zames Chew started Repair.sg with his 14-year-old brother, Amos Chew.

No training. No mentors. No blueprint.

They learned repairs by shadowing handymen they found on classifieds — even selling leads just to tag along.

🇸🇬 For years, Zames answered customer inquiries at 4AM and drove out to do repairs by 7AM. 7 days a week. No systems. No backup.

The business was at the brink of death for 7 years straight.

Then COVID hit. Jobs vanished. The loan they took to grow ended up being used just to survive.

Everything felt like it could collapse at any moment.

But that lockdown became the best thing for the business.

With no jobs to rush to, Zames and Amos finally had time to think. They started hiring and delegating. Let go of team members who didn't fit. Rebuilt the culture from scratch.

2024 — S$1.7M revenue. 20,000+ customers.
2025 — S$3M+ revenue. HaiDiLao and Coffee Bean as clients.

All bootstrapped. Zero funding raised.

Tonight at 8pm, I'm publishing the full story — how two teenage brothers built one of Singapore's most trusted handyman brands, the 2 mistakes that nearly killed them, and the exact changes that turned everything around.

Subscribe to read it → https://lnkd.in/gdDrhgtq

P.S. S$30 to S$3M. No funding. Just grit. I sat down with Zames for a full podcast — he held nothing back. Story + podcast drops tonight.

Admond Lee

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

$23M raised. 300,000 users. Licensed in Malaysia.

Less than 1% of customer funds left.

🇸🇬 Tokenize Xchange looked like the real deal — Singapore's most promising crypto exchange.

Founded by Hong Qi Yu in 2017, TX had everything going for it:
- $23M in Series A funding
- Full regulatory license in Malaysia
- 300,000+ verified users across SEA
- S$266M in customer assets under custody
- Kenanga Investment Bank as a shareholder

Then MAS rejected their license application.

Within weeks:
- Founder arrested and charged with fraud
- Withdrawals frozen overnight
- S$266M owed to 2,200+ customers
- Only S$2.6M in available assets

That's less than 1% to cover what they owed.

272 users are now suing the founder for S$60M+.

I broke down the full story — from $1B token peak to fraud charges.

Read it here → https://lnkd.in/gZq4xgTW

P.S. S$266M owed. S$2.6M left. That ratio still blows my mind.

S$266M in customer assets. S$2.6M left.

That's less than 1% of what Tokenize Xchange owed its users.

🇸🇬 In 2017, Hong Qi Yu was a frustrated crypto trader in Singapore. He couldn't find a reliable platform to buy Bitcoin with SGD — delays, clunky UIs, no proper order books. So he built Tokenize Xchange to fix it.

At its peak:
- $23M raised in Series A
- 300,000+ verified users
- Licensed in Malaysia
- TKX token hit $1B market cap

Then MAS rejected their license in July 2025.

Within weeks:
- Withdrawals frozen
- Founder arrested and charged with fraud
- S$266M in user funds vanished
- 2,200+ customers left with nothing

The judicial managers found only S$2.6M in available assets. Customer funds were allegedly commingled with company operating capital.

272 users have now filed a lawsuit for over S$60M in damages.

Tonight at 8pm, I'm publishing the full story — the timeline, the 2 fatal mistakes, and what every founder and crypto user should learn from this collapse.

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

P.S. He built Tokenize Xchange because he wanted a safer way to buy crypto. The story didn't end that way.

$1.3B revenue. 50,000 employees. Supplying NATO.

CEO arrested. 10,665 workers laid off.

🇮🇩 Sritex started as a tiny textile stall in Solo, Indonesia, 1966.

H.M. Lukminto wasn't manufacturing anything — just buying and selling cloth. But he quickly realised trading margins were thin and quality was outside his control.

So he built his own factory. Then another. Then another.

By 2019, Sritex had become Southeast Asia's largest vertically integrated textile empire:
- $1.3B in annual revenue
- 50,000+ workers across multiple factories
- Supplying military uniforms to 35 countries (including NATO)

But all of it was built on borrowed money.

$500M in USD bonds. Fixed costs that didn't care if global demand vanished overnight.

Then COVID hit.

The same scale that made Sritex powerful became its biggest burden. Restructuring failed. Operations ceased. 10,665 workers lost their jobs.

And in May 2025, the former CEO was arrested on corruption charges.

Tonight at 8pm, I'm publishing the full story — how a market stall became a $1.3B empire, the 2 mistakes that brought it down, and what founders can learn from one of Southeast Asia's biggest corporate collapses.

Subscribe to read it → https://lnkd.in/gdDrhgtq

P.S. Market stall to NATO supplier to bankruptcy. 59 years of building, gone.

8 years on LinkedIn. I've burned out more times than I can count.

If you're a content creator here, you know what I mean.

Some weeks you're on fire. Others, you're staring at a blank screen wondering why you even bother.

What keeps me going? Moments like this.

CNY lunch with a bunch of LinkedIn friends. We sat around a table, ate good food, and laughed like we've known each other for years.

We all met on LinkedIn first. And now we're sharing lo hei like old friends.

Not connections. Not networking. Actual friendships.

8 years of posting. 
The algorithm changes. 
The engagement fluctuates. 
The burnout hits.

But the friendships compound.

Thank you Vincent PHANG for bringing us together. And to everyone in this photo — grateful our paths crossed on this app.

Elliot Lee, PhD, Adrian Kee 🍀, Judson Teo, Kelvin Kuok, Elizabeth Loh, Sam Neo, Eugene Ng, Jun Yi Lee ↗️ 

💪🏻

I rejected a founder from Beyond Runway yesterday.

She reached out to join. Seemed excited. Had a clear vision.

Then I asked about revenue.

"We're still pre-revenue."

I had to say no.

It wasn't personal. It was the standard we set from day one.

Beyond Runway is a private community for founders doing at least $100K annual revenue. No exceptions.

Why the bar?

Because most founder communities suck. They're full of:
- Promo spam and humble-brags
- Non-founders or hobby businesses
- Generic advice that doesn't apply to your stage
- Too big, too noisy, no curation

When you're doing $100K+, you don't need motivation. You need specific answers from founders 1-2 steps ahead.

So I built something different.

Inside, you'll find VC/YC-backed founders, bootstrapped operators doing $5M+, and everyone in between — all vetted, all building real businesses.

What you get:
• Private WhatsApp group with 20+ vetted founders ($100K–$20M revenue)
• Monthly masterclasses with top operators
• Direct access to me for Q&A
• Warm intros to founders and investors when it makes sense (I've lost count of the number of warm intros I've done lol)

We're capping at 50 founders, because small rooms have higher signal.

If you're doing $100K+ and want a tight circle of serious operators who share what actually works, this is it.

Join here → https://lnkd.in/gzYZj_3t

P.S. To the founder I rejected — keep building. When you hit $100K, the door's open.

HK$30 million of personal savings.

Still couldn't save a 33-year-old business.

🇭🇰 Ocean Empire Food Shop was Hong Kong's most iconic congee chain — founded in 1992 by two partners determined to modernise congee.

Congee was beloved. But the experience? Cramped dai pai dong stalls, inconsistent hygiene, unreliable quality.

They wanted to fix that.

At its peak:
- 30+ outlets across Hong Kong and Macau
- 16,000 bowls sold per day
- ISO 22000 and HACCP certified
- Founder awarded the Silver Bauhinia Star (SBS)

From street food to city-honoured institution.

Then 2019 hit (protests). Then COVID.

The founders didn't walk away. They injected HK$30M+ of personal funds. Sold property. Took loans. Did everything to keep the business alive.

It wasn't enough.

By 2025:
- Branches shrank from 30+ to 7
- Rent arrears piled up
- Employees chased unpaid wages
- Founder declared bankrupt

33 years of building. Gone.

Tonight at 8pm, I'm publishing the full story — why pouring HK$30M of personal money couldn't save a business that was already structurally broken.

Subscribe to read it → https://lnkd.in/gdDrhgtq

$58M raised. 300 employees. Fortune 500 clients.

12 years later, she had to sell everything.

Rosaline quit her corporate career at 50+. Put $10M of her own money on the line — $5M savings, $5M debt. Started CXA from her living room.

Her goal? Build a unicorn before 60.

At its peak:
• 600+ enterprise clients including Fortune 500s
• 400,000 employees served across 20 countries
• 50% annual revenue growth
• Break-even in sight

Then it all unraveled. Slowly.

CXA didn't explode. It bled out — layoffs, asset sales, hub shutdowns, pivot after pivot.

⚠️ The fatal mistake: running 2 businesses at once.

A regulated, people-heavy brokerage AND a capital-intensive SaaS platform. Both demanding. Both draining. Neither sustainable together.

Back then, building software was expensive. You needed large dev teams across multiple countries just to test if an idea worked. CXA had 70 engineers in Vietnam alone.

By the time they realised the SaaS model wasn't proven, they'd already burned through millions.

That's the part that stuck with me.

A friend of mine spent $20k on a dev agency. Took them a year to build an MVP. It didn't even work.

Today, founders don't need 70 engineers — or dev agencies — to validate a SaaS idea. Tools like Replit let you prototype in a weekend, test with real users, and know if something works before burning through your runway.

What took millions to build back then, you can now test for almost nothing.
If you're building something, validate cheap. Validate fast.

Try Replit → https://lnkd.in/gwEWVKAt

P.S. Rosaline fought for 12 years. Her biggest lesson? She wasn't frugal enough early on.

undefined undefined

Replit

1.5 years. $0 revenue. 2 months of runway left.

That was Staq — my first startup.

We built a universal API for digital lenders to access SME financial data. Months of user interviews. Months of building. Pilots with big lenders. Sandbox integrations.

On paper, we had momentum.

In reality? Zero paid contracts. Cash bleeding out. No time left to iterate.

By December 2022, we were dying. Our last resort was an acquire-hire deal. We waited 4 months in limbo. No clarity. No backup plan. Just waiting to hear if we still had a future.

April 2023 — the deal fell through.

Staq was dead. That period broke me.

⚠️ 2 mistakes that killed us:
1. No founder-market fit — we had no domain expertise, no network. We spent months validating what insiders would've known in weeks.

2. Fell in love with the solution — we copied Codat's UK model without deeply understanding the problem in SEA. By the time we realised, we had no runway left to pivot.

The truth? We ran out of time before we could find something that worked.

If I were starting Staq today, I'd use Replit.

Not because it would've saved us. But because it would've let us test 10 ideas in the time we spent building 1. Spin up an MVP in days, not months. Validate with real users before burning runway.

What took us 3 months to build in 2022, founders can now ship in a weekend.

If you're building something, move fast. Seriously.

Try Replit → https://lnkd.in/gwEWVKAt

P.S. Staq failed. But if my story helps one founder move faster and validate sooner, it was worth sharing.

YC just dropped 10 startup ideas they want to fund.

One of them hits close to home: AI-Native Agencies.

I ran an agency before — learned this the hard way.

Agencies are brutal to scale.
- Low margins
- Slow manual work
- The only way to grow? Hire more people.

You trade time for money. Forever.

But AI flips the model.

Instead of selling software to help customers do the work, you use the software yourself and sell them the finished product — at 100x the price.

Think about it:
- A design firm that delivers custom work before the contract is signed
- An ad agency that creates video ads without physical shoots
- A law firm that writes legal docs in minutes, not weeks

Agencies of the future will look more like software companies. With software margins.

A few other ideas from YC's list that caught my eye:
- Cursor for Product Managers
- AI-Native Hedge Funds
- Infra for Government Fraud Hunters
- Large Spatial Models

Full list here: https://lnkd.in/gvY_gCFK

P.S. Anyone here running an agency thinking about this shift? Curious how you're using AI to change the model.

Most founders pitch investors wrong. He raised $11.2M doing the opposite.

Himanshu Upreti is the co-founder & CTO of Ai Palette — acquired by GlobalData in 2025 after raising $11.2M.

I invited him to our Beyond Runway masterclass. Here are 10 takeaways:

💰 BEFORE YOU RAISE

• Validate a "hair-on-fire" problem first. Use The Mom Test. Get to the truth, even if it hurts.

• Paid pilots > free POCs. Even $10-$20 signals real value. Free means they don't care.

💰 GETTING YOUR FIRST TERM SHEET

• Find one anchor investor. Get their term sheet. Use it to trigger FOMO with others.

• Tier your investor list. Pitch Tier 3 first (practice). Tier 2 next (warm). Tier 1 last (dream investors). By then, you'll have answers to every question.

💰 RUNNING THE PROCESS

• Treat fundraising like B2B sales. Build a CRM. Research each VC's thesis. Personalise every message based on their portfolio.

• Warm intros beat cold emails. Founder-to-founder referrals convert way higher.

• Traction > narrative. Investors didn't understand AI for CPG. But they saw Unilever and Kellogg's paying.

💰 AFTER YOU RAISE

• Monthly updates. Quarterly board meetings. Don't overpromise.

• Grab coffee. Build human connections beyond the cap table.

💰 TIMING

• Raise when you have 12-18 months runway left. Desperation = bad valuations.

• Raise just enough. 15 focused people can outperform a 30-person team.

What's one fundraising tip you'd add to this list?

P.S. This masterclass was exclusive to Beyond Runway — a private community for revenue-generating founders who learn fast, move faster, and win together. Join us → https://lnkd.in/gzYZj_3t

Co-Founder at AI Palette | Forbes 30 under 30 | Ex-Visa, IIT | AI Speaker, Angel Investor

Ai Palette

$25M raised. 200,000 users. Collapsed overnight.

Style Theory had everything — SoftBank backing, Orchard Road flagship, 1 million rentals delivered.

Revenue was growing. Users were scaling.

But here's what the numbers didn't show:

Every single rental lost money.

Two-way logistics ate them alive. Ship out, clean, repair, ship back. The more customers they got, the faster they bled cash.

200,000 users sounds impressive — until you realise each user cost more to serve than they paid.

September 30, 2025. Operations ceased. No warning. No wind-down.

The founders weren't reckless. They just didn't have financial clarity until it was too late.

I've written 100+ startup autopsies. This is one of the most common killers — not bad products, not weak teams. Just founders who didn't know their real numbers until the bank account hit zero.

OSOME is hosting a free 30-min webinar on 10 Feb (Tuesday), 5PM — on getting real financial clarity without spreadsheets or jargon.

What you'll learn:
• The 3 reports that actually tell you if you can afford that next hire
• How to stop spending money that doesn't exist yet
• Key compliance deadlines you can't afford to miss

Capped at 50 founders. Register here → https://lnkd.in/gE2KvNVT

P.S. The graveyard is full of startups that looked unstoppable on paper.

Airwallex just made traditional banking feel like fax machines.

A few days ago, they launched their AI Assistant.

I opened a USD account in 4 seconds.

Not 4 days. Not 4 hours. 4 seconds.

I was testing it — and honestly, I wasn't expecting much.

Most "AI assistants" are glorified FAQ bots. Ask something slightly complex and they break.

This one didn't.

So I tried it:

→ Told it I needed to receive USD from customers — helped me open a USD account instantly
→ Wanted corporate cards to categorise my business expenses — set it up in minutes
→ Confused about the dashboard — walked me through step by step
→ Wanted to check my revenue and spend — pulled the breakdown instantly
→ Had a payment tracking issue — auto-escalated to human support without me asking

That last one surprised me most.

It knew when it couldn't solve something — and handed it off seamlessly. No dead ends. No "please contact support" loops.

No more clicking through 10 screens or digging through docs. Just ask, get it done, move on.

Finally, an AI that actually gets stuff done.

Try for yourself → https://lnkd.in/gvwbNc5Y

P.S. Traditional banks: "Please allow 3-5 business days." Airwallex: "Done."

S$30. No training. No connections.

Two teenage brothers built a S$3M handyman empire.

🇸🇬 In 2016, Zames Chew was 16. His parents needed something fixed at home, but finding a reliable repairman online was nearly impossible.

So he bought a domain (Repair.sg) with his savings and registered the business with his dad's help — because he was underage.

His 14-year-old brother Amos Chew joined him. Neither had any formal training.

To learn the trade, they found veteran handymen through classifieds and sold them leads just to tag along and shadow their work.

For 7 years, the business was at the brink of death most of the time.

Zames answered customer inquiries at 4am and drove out to do repairs at 7am. They made every mistake — over-investing in logos instead of service quality, hiring without culture, doing everything themselves until burnout.

The lowest point? COVID lockdown. A loan they took to grow ended up being used just to survive.

Today: 20+ staff. S$3M+ revenue. Clients like HaiDiLao and Coffee Bean.

How did two burned-out teenage brothers turn a dying business into Singapore's most trusted handyman brand?

Their full story and podcast is out → https://lnkd.in/gtWVkn6F

P.S. Every time Zames sees another yellow Repair.sg van on the road, he knows that S$30 gamble was worth it.

CEO @ Repair.sg | Fastest-Growing Property Maintenance Company for Businesses and Homeowners with 1800+ Google Reviews | Loved By More Than 30,000 Happy Customers! ❤️

COO @ Repair.sg. Taking care of your property!

Repair.sg

You don't need to lose $10M to learn a $10M lesson.

Most startups fail for the exact same reasons.

You can spend 5 years and your life savings learning those lessons the hard way — or you can spend 5 minutes every Sunday getting the post-mortem for free from The Runway Ventures.

Every week, I break down 1 failed startup story from Asia:

🧠 1 Failed Startup Story — the raw, unfiltered truth
⚠️ 2 Fatal Mistakes — exactly where they went off the rails
🛠️ 3 Actionable Lessons — frameworks you can use today to de-risk your business

No theories. No fluff. Just the signals you need to stay alive and scale.

I've covered:
• CXA — $58M raised, 12 years of building, sold everything
• Style Theory — $25M raised, 200k users, collapsed overnight
• Gravel — $14M raised, shut down exactly 1 year later

20,000+ founders already get this every Sunday.

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

P.S. 100+ startup autopsies and counting. Every mistake in here is one you won't have to make.

$1.3B revenue. 50,000 employees. NATO supplier.

CEO arrested on corruption allegations. 10,000+ workers laid off.

🇮🇩 Sritex was Southeast Asia's largest textile empire.

It started as a tiny market stall in Solo, 1966. H.M. Lukminto was just buying and selling cloth.

By the 2010s, Sritex had become a vertically integrated powerhouse — spinning, weaving, dyeing, and garment production all under one roof.

At its peak:
- $1.3B annual revenue
- 50,000+ employees
- Exported to 55 countries
- Supplied military uniforms to 35 countries (including NATO)

Then they issued $500M in USD bonds to scale even faster.

COVID hit. Global demand collapsed. The debt crushed them.

Restructuring failed. Courts stepped in. Operations ceased in March 2025.

Then came the corruption charges.

The former CEO was arrested in May 2025 over allegations tied to IDR 1.35 trillion in loans.

From market stall to empire to collapse — in one generation.

I broke down the full story, the 2 fatal mistakes, and what founders can learn from this.

Read it here → https://lnkd.in/g5NnMXNj

P.S. 59 years to build. 4 years to collapse.

Biofourmis is just the tip of the iceberg. Trust me. The whistleblower's insider news that I've received is even crazier. Our ecosystem is recovering, yet there are still many bad actors lurking behind the scenes. 

God bless SEA.

P.S. If anyone wants to share more about Biofourmis with me, my DM is open. Looking to cover the story soon.

I make learning Data Science effortless for you | Data Science @ Bosch

Hi 👋🏻
I'm Banias, AI Engineer at Netze BW.

Applying new technologies and methods in Data Science & AI is my passion.

Through my experience and continual learning, I've gained valuable insights and tips that I love sharing with Data Scientists, Machine Learning Engineers and builders like you.

If you're looking for practical advice, helpful resources, and articles on the latest trends or best practices in the field, you're in the right place.

Let's connect and stay updated on the exciting world of Data Science together. Feel free to reach out, read my articles, and let's grow our knowledge and network. :)


Note: My views are my own and don‘t represent my employer.

Banias Baabe

Running a 70B LLM on a single 4GB GPU is possible with 𝗔𝗶𝗿𝗟𝗟𝗠.

𝗔𝗶𝗿𝗟𝗟𝗠 optimizes inference memory usage so you can run big models without quantization, distillation, or pruning.

It runs inference layer by layer, loading only the layer that is needed for the next compute step, then freeing that memory before moving on.​

To make that practical, it first preprocesses the original Hugging Face weights and re-shards them by transformer layer (instead of giant multi‑GB shards), so each load is focused and smaller.​

This matters because disk I/O becomes the bottleneck when you stream weights.

The main thing that still sits in GPU memory is the KV cache for attention, which is much smaller than full model weights for typical context lengths.​

𝗔𝗶𝗿𝗟𝗟𝗠 also adds prefetching so it can overlap loading the next chunk of weights with the current compute, which helps claw back some speed.

And if you want more throughput, it supports an optional “model compression” mode using block-wise quantization of weights to shrink what needs to be loaded from disk, with the project claiming up to ~3x speedup and minimal accuracy loss.

🔗 Link to repo: github(.)com/lyogavin/airllm


---
♻️ Found this useful? Share it with another builder.
➕ For daily practical AI and Python posts, follow Banias Baabe.

I teach AI Engineers the concepts and tools they need to stay ahead | AI Engineer @ Netze BW

Tired of blocking I/O messing up your async Python apps?

Try 𝗮𝗶𝗼𝗳𝗶𝗹𝗲𝘀 - the game-changer for handling files in asyncio without the headaches.

It's an Apache-licensed library that wraps standard file ops into async magic.

Why you should use it:
• Mirrors Python's built-in file API, so it's super familiar.
• Supports buffered text, binary files, and even tempfiles asynchronously.
• Delegates to a thread pool to keep your event loop flying.


🔗 Link to repo: github(dot)com/Tinche/aiofiles

---
♻️ Found this useful? Share it with another builder.
➕ For daily practical AI and Python posts, follow Banias Baabe.

Do you want to identify which lines use the most amount of memory in your Python script?

Try 𝗺𝗲𝗺𝗼𝗿𝘆_𝗽𝗿𝗼𝗳𝗶𝗹𝗲𝗿.

𝗺𝗲𝗺𝗼𝗿𝘆_𝗽𝗿𝗼𝗳𝗶𝗹𝗲𝗿 is a Python module to make a line-by-line analysis of memory consumption for Python functions.

Below you can see how to use memory_profiler within your Python script:

• Decorate the function you want to profile with @𝗽𝗿𝗼𝗳𝗶𝗹𝗲.
• Run the script by passing the option -𝗺 𝗺𝗲𝗺𝗼𝗿𝘆_𝗽𝗿𝗼𝗳𝗶𝗹𝗲𝗿 to load the 𝗺𝗲𝗺𝗼𝗿𝘆_𝗽𝗿𝗼𝗳𝗶𝗹𝗲𝗿 module.

Link to project: pypi(dot)org/project/memory-profiler


---
♻️ Found this useful? Share it with another builder.
➕ For daily practical AI and Python posts, follow Banias Baabe.

LLMs have a goldfish memory problem.

But 𝗠𝗲𝗺𝗼𝗿𝗶 could be the solution.

It’s an open-source, SQL-native memory engine that gives your agents persistent long-term memory with one single line of code.

Here’s why it's interesting:

✅ SQL-Native: No vector DB required. Use the Postgres, MySQL, or SQLite you already have.
✅ Framework Agnostic: Works seamlessly with OpenAI, Anthropic, LangChain, and Pydantic AI.
✅ Advanced Augmentation: Automatically extracts entities, facts, and user preferences in the background.
✅ Zero Latency: The new threaded architecture means memory storage doesn't slow down your chat.
✅ Graph-Ready: It now stores data in third normal form with semantic triples, effectively building a knowledge graph for free.


---
♻️ Found this useful? Share it with another builder.
➕ For daily practical AI and Python posts, follow Banias Baabe.

Struggling with slow data processing for AI training?

Try 𝗹𝗶𝘁𝗱𝗮𝘁𝗮, the Python library that transforms and optimizes datasets at scale!

𝗹𝗶𝘁𝗱𝗮𝘁𝗮 offers:
🟢 Streaming large cloud datasets without local downloads
🟢 Accelerating training by 20x with optimized data
🟢 Parallelizing tasks across multiple machines effortlessly
🟢 Securely handling sensitive data with advanced encryption

Perfect for images, text, video, and more.

Check it out: github(dot)com/Lightning-AI/litdata


---
♻️ Found this useful? Share it with another builder.
➕ For daily practical AI and Python posts, follow Banias Baabe.

taplio

The Battle

Admond Lee Kin Lim

Banias Baabe

Followers

Posts

Engagements

Likes

Performance ScoreAdmond Lee Kin Lim's LinkedIn Performance

Performance ScoreBanias Baabe's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Admond Lee Kin Lim

Banias Baabe

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 ScoreBanias Baabe's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreAdmond Lee Kin Lim's LinkedIn Performance