https://x.com/andrewfarah/status/2040535592925372895?s=46
Try this out if you are using X:
—
Some ideas:
1. Based on all the articles, news, and links I’ve bookmarked on X, how can they help with my current work or personal tasks? For example, I’m working on [briefly mention your main task, e.g. preparing a report for my boss / planning my child’s tuition / researching resale HDB flats / improving my sales targets]. Suggest 3–5 practical ideas.
2. Looking at the topics and links I’ve saved on X, what 3–5 books, would you recommend I read or learn next? Focus on things that can help me in my daily life in Singapore, like career growth, parenting, health, property, or managing stress.
3. I’ve bookmarked many articles and studies about biohacking, price predictions, etc. Based on what I saved, can you summarise what actually happened with those studies or findings later?
- I often bookmark articles about [your common topics, e.g. CPF planning, food delivery business ideas, fitness for office workers, or raising kids] What patterns do you see in my interests, and what should I focus on or learn next?
- Summarise the key takeaways from all the property / investment / parenting / career articles I tend to save on X. Give me simple action steps suitable for someone living in Singapore
Try this out if you are using X:
—
Some ideas:
1. Based on all the articles, news, and links I’ve bookmarked on X, how can they help with my current work or personal tasks? For example, I’m working on [briefly mention your main task, e.g. preparing a report for my boss / planning my child’s tuition / researching resale HDB flats / improving my sales targets]. Suggest 3–5 practical ideas.
2. Looking at the topics and links I’ve saved on X, what 3–5 books, would you recommend I read or learn next? Focus on things that can help me in my daily life in Singapore, like career growth, parenting, health, property, or managing stress.
3. I’ve bookmarked many articles and studies about biohacking, price predictions, etc. Based on what I saved, can you summarise what actually happened with those studies or findings later?
- I often bookmark articles about [your common topics, e.g. CPF planning, food delivery business ideas, fitness for office workers, or raising kids] What patterns do you see in my interests, and what should I focus on or learn next?
- Summarise the key takeaways from all the property / investment / parenting / career articles I tend to save on X. Give me simple action steps suitable for someone living in Singapore
Some observations over the past 2 weeks:
Multi-agent setups with OpenClaw & Hermes
• OpenClaw handles execution, while Hermes performs QC, checking for errors before OpenClaw drifts and suggesting improvements.
It’s essentially a self-improving loop.
Personally I’ve been using GPT-5.4 to QC Opus’s code 😂
Different models pick up on different issues.
Mostly experimenting for now. I’m not sure if it’s worth the extra compute unless it’s for a project that justifies it.
Multi-agent setups with OpenClaw & Hermes
• OpenClaw handles execution, while Hermes performs QC, checking for errors before OpenClaw drifts and suggesting improvements.
It’s essentially a self-improving loop.
Personally I’ve been using GPT-5.4 to QC Opus’s code 😂
Different models pick up on different issues.
Mostly experimenting for now. I’m not sure if it’s worth the extra compute unless it’s for a project that justifies it.
👍2
Tuesday morning:
exhausted my $30 ChatGPT Codex plan, and claude is too precious for research, so rotating back to Gemini for some “planning” 😂
Use tip: Ask your LLM to filter through links/bookmarks you see online - are they LARPing or engagement bait - which ones are actually usable. You will save yourself a LOT of time. Your final filter should be your own sense of judgement.
You can then proceed to unfollow/mute those accounts to keep your signal noise low
exhausted my $30 ChatGPT Codex plan, and claude is too precious for research, so rotating back to Gemini for some “planning” 😂
Use tip: Ask your LLM to filter through links/bookmarks you see online - are they LARPing or engagement bait - which ones are actually usable. You will save yourself a LOT of time. Your final filter should be your own sense of judgement.
You can then proceed to unfollow/mute those accounts to keep your signal noise low
👍3😁1🤣1
How someone used Openclaw for his Dad’s small tea business (worth a read!)
• Automates supplier orders, print packing slips and order PDF every 5 minutes
• Built website pages for Dad
• Tracks shipments w/ daily updates, folllow up on orders, calculates payrolls, etc
• All running on Telegram (!!!)
https://x.com/danpeguine/status/2039330885917442238
• Automates supplier orders, print packing slips and order PDF every 5 minutes
• Built website pages for Dad
• Tracks shipments w/ daily updates, folllow up on orders, calculates payrolls, etc
• All running on Telegram (!!!)
https://x.com/danpeguine/status/2039330885917442238
X (formerly Twitter)
Dan Peguine ⌐◨-◨ (@danpeguine) on X
How a small business owner is really using OpenClaw (2 months in)
❤8🤩1
Claude’s Mythos hitting the timeline,
They are claiming that it has found thousands of high-severity vulnerabilities for some major OS/browsers.
Not sure how to feel about this.
https://x.com/AnthropicAI/status/2041578403686498506
They are claiming that it has found thousands of high-severity vulnerabilities for some major OS/browsers.
Not sure how to feel about this.
https://x.com/AnthropicAI/status/2041578403686498506
X (formerly Twitter)
Anthropic (@AnthropicAI) on X
Mythos Preview has already found thousands of high-severity vulnerabilities—including some in every major operating system and web browser.
I tried Karpathy’s llm-wiki on my own saved PDFs, and the hardest part wasn’t the model.
It was getting the workflow to fit how I actually consume information.
Original idea:
https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f
https://x.com/karpathy/status/2039805659525644595
I tested it on a few PDF books/articles I’d been sitting on, mostly around wellness/fitness.
What stood out quickly: the value wasn’t just “upload documents, get summaries.”
The real value was in everything around that:
ingestion, extraction, structure, and turning the output into something reusable.
A few things were messier than expected:
• I first used Gemini to help think through prompts and structure, but that didn’t map cleanly to what Codex actually needed (usage limits lol)
• The PDFs weren’t plug-and-play either. Extraction/readability became their own problem. I used some critical thinking to look at what plugins Hermes/Openclaw used for PDFs. I needed it to OCR and read text/images. The basic ones Codex recommended wouldn’t cut it
• Then I realized the bigger issue was my ingestion pipeline. I don’t just want this for PDFs / simple notes on obsidian. I want it to work with images, Telegram messages, and X links too.
What worked:
once I started debugging and using superpowers w/ Codex, the flow started working, I got enriched, structured notes instead of passive highlights. The output actually felt reusable, less like an “AI summary” and more like a knowledge base that I’d actually keep using.
Big takeaway:
• copy-pasta Karpathy’s model was the easy part, getting the workflow to fit my actual inputs was the real work.
• Also a good reminder that things always look cleaner on X than they do when you try to make them work in your own setup.
Curious if anyone else here has tried building a personal workflow like this around their saved material?
It was getting the workflow to fit how I actually consume information.
Original idea:
https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f
https://x.com/karpathy/status/2039805659525644595
I tested it on a few PDF books/articles I’d been sitting on, mostly around wellness/fitness.
What stood out quickly: the value wasn’t just “upload documents, get summaries.”
The real value was in everything around that:
ingestion, extraction, structure, and turning the output into something reusable.
A few things were messier than expected:
• I first used Gemini to help think through prompts and structure, but that didn’t map cleanly to what Codex actually needed (usage limits lol)
• The PDFs weren’t plug-and-play either. Extraction/readability became their own problem. I used some critical thinking to look at what plugins Hermes/Openclaw used for PDFs. I needed it to OCR and read text/images. The basic ones Codex recommended wouldn’t cut it
• Then I realized the bigger issue was my ingestion pipeline. I don’t just want this for PDFs / simple notes on obsidian. I want it to work with images, Telegram messages, and X links too.
What worked:
once I started debugging and using superpowers w/ Codex, the flow started working, I got enriched, structured notes instead of passive highlights. The output actually felt reusable, less like an “AI summary” and more like a knowledge base that I’d actually keep using.
Big takeaway:
• copy-pasta Karpathy’s model was the easy part, getting the workflow to fit my actual inputs was the real work.
• Also a good reminder that things always look cleaner on X than they do when you try to make them work in your own setup.
Curious if anyone else here has tried building a personal workflow like this around their saved material?
❤3
Someone built a real-time AI tutor that looks at your screen and cursor, talks to you, and points at stuff answering questions that you have.
All open-source and free.
In the tutorial he’s asking how to use a video editing app. Pretty wild.
I can see a lot of use cases happening with this tool. So far:
Examples of usage so far:
- A Mom building her first app on Lovable
- A dentist debugging his OpenClaw setup
- A photographer getting feedback in Lightroom
- Founders keeping track of their todos.
- Designers getting feedback in Figma
- A student outlining her thesis in G-Docs
- Traders analyzing live stock charts
Try it out!
https://x.com/farzatv/status/2041314633978659092?s=46
All open-source and free.
In the tutorial he’s asking how to use a video editing app. Pretty wild.
I can see a lot of use cases happening with this tool. So far:
Examples of usage so far:
- A Mom building her first app on Lovable
- A dentist debugging his OpenClaw setup
- A photographer getting feedback in Lightroom
- Founders keeping track of their todos.
- Designers getting feedback in Figma
- A student outlining her thesis in G-Docs
- Traders analyzing live stock charts
Try it out!
https://x.com/farzatv/status/2041314633978659092?s=46
X (formerly Twitter)
Farza 🇵🇰🇺🇸 (@FarzaTV) on X
I built this thing called Clicky.
It's an AI teacher that lives as a buddy next to your cursor.
It can see your screen, talk to you, and even point at stuff, kinda like having a real teacher next to you.
I've been using it the past few days to learn Davinci…
It's an AI teacher that lives as a buddy next to your cursor.
It can see your screen, talk to you, and even point at stuff, kinda like having a real teacher next to you.
I've been using it the past few days to learn Davinci…
❤1
The AI Burrow 🐰🕳️
I tried Karpathy’s llm-wiki on my own saved PDFs, and the hardest part wasn’t the model. It was getting the workflow to fit how I actually consume information. Original idea: https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f https://x.com…
We live in exciting times. Built myself a 3rd reiteration of my second brain 🧠 Turning years of scattered thoughts/unread books, links, articles into a wiki that actually teaches itself. The more raw data I feed it, the more it self improves, finding connections between my interests that I never even noticed.
👍6
https://luma.com/srcfgry1
Grab is open in up their grabmaps API and inviting people to tinker with it in a one day
- early access to build with tools nobody else has yet.
- elevated API credits, hands-on support from grab engineers,
- 7 hours of hacking time, and over $6,000 in prizes.
three tracks:
- intelligent mobility - dynamic routing, real-time delivery, ETA tools
-discover the city - hyperlocal recs, social maps, curated experiences
- smarter places trending locations, richer POl data, activity visualisation.
all skill levels welcome
Anyone here keen?
Grab is open in up their grabmaps API and inviting people to tinker with it in a one day
- early access to build with tools nobody else has yet.
- elevated API credits, hands-on support from grab engineers,
- 7 hours of hacking time, and over $6,000 in prizes.
three tracks:
- intelligent mobility - dynamic routing, real-time delivery, ETA tools
-discover the city - hyperlocal recs, social maps, curated experiences
- smarter places trending locations, richer POl data, activity visualisation.
all skill levels welcome
Anyone here keen?
❤1
My Recap from Carousell’s Agentic @ Work
Went to a small but packed meetup last night with practitioners who are actually shipping AI in production. A few things stuck with me.
Airwallex (Global fintech company) These guys have gone all in on AI.
• Their Engineers no longer write code, they supervise AI agents now. They have integrated Claude into their own company LLM and use it for their entire workflow
Their agent reads the brief, writes the code, opens the pull request.
One interesting takeaway was that in the future, your Token usage will show up in performance reviews
GovTech / Career Kaki
Singapore’s government tech team building for everyday Singaporeans.
- AI career coach, log in with Singpass and it already knows your background, connected to OpenAI/Gemini
The problem they’re really solving: most people don’t seek career help until it’s too late.
The theme across every talk was the same. AI isn’t coming. It’s already inside how teams operate. The gap between companies using AI and those still thinking about it is widening fast.
Went to a small but packed meetup last night with practitioners who are actually shipping AI in production. A few things stuck with me.
Airwallex (Global fintech company) These guys have gone all in on AI.
• Their Engineers no longer write code, they supervise AI agents now. They have integrated Claude into their own company LLM and use it for their entire workflow
Their agent reads the brief, writes the code, opens the pull request.
One interesting takeaway was that in the future, your Token usage will show up in performance reviews
GovTech / Career Kaki
Singapore’s government tech team building for everyday Singaporeans.
- AI career coach, log in with Singpass and it already knows your background, connected to OpenAI/Gemini
The problem they’re really solving: most people don’t seek career help until it’s too late.
The theme across every talk was the same. AI isn’t coming. It’s already inside how teams operate. The gap between companies using AI and those still thinking about it is widening fast.
🔥6❤1
Enjoy the $20 - $200 all you can use monthly subscrptions while you can …
Anthropic cutting off OpenClaw access is an early warning sign that the $200/month Claude Max and ChatGPT Pro “unlimited” plans won’t last.
TLDR: The economics don’t work, Intelligence/Compute will get more expensive.
I foresee by the end of year $100/month subscriptions being the baseline norm for these SOTA models.
https://creatoreconomy.so/p/the-all-you-can-use-ai-subscription
Anthropic cutting off OpenClaw access is an early warning sign that the $200/month Claude Max and ChatGPT Pro “unlimited” plans won’t last.
TLDR: The economics don’t work, Intelligence/Compute will get more expensive.
I foresee by the end of year $100/month subscriptions being the baseline norm for these SOTA models.
https://creatoreconomy.so/p/the-all-you-can-use-ai-subscription
creatoreconomy.so
The All-You-Can-Use AI Subscription Won't Last Forever
Thoughts on the Anthropic and OpenClaw drama, running local models, and what I learned about AI on the ground in China
❤2
https://x.com/basedonex/status/2042546883411779841?s=46
PSA: CLAUDE MAX / CHATGPT PRO REBATES
If you are not a Singapore citizen/from any of the non-excluded country list, you can use the based card to get free ChatGPT/Claude Max sub (up to $100usd) rebated on the based gold card.
PSA: CLAUDE MAX / CHATGPT PRO REBATES
If you are not a Singapore citizen/from any of the non-excluded country list, you can use the based card to get free ChatGPT/Claude Max sub (up to $100usd) rebated on the based gold card.
X (formerly Twitter)
Based (@BasedOneX) on X
Your Claude and OpenAI subscriptions are now free on Based Cards.
Gold Card holders get a free ChatGPT Pro or Claude Max plan every month, up to $100 rebated. HYPE Card holders get $10 rebated on the same spend.
The tools you already pay for, auto-credited…
Gold Card holders get a free ChatGPT Pro or Claude Max plan every month, up to $100 rebated. HYPE Card holders get $10 rebated on the same spend.
The tools you already pay for, auto-credited…
Good read, especially with all the AI stimuli.
Spoke to some friends who feel they are very left behind because of how fast things are. The reality is still we are in an echo chamber. The masses still have not caught on yet. We think we are lagging behind because of the algo we are being fed.
Modern life feels dull because our brains are getting trained by “superstimuli” - social media, hyper-edited content, instant novelty, vanity metrics, junk food, etc. These hit our reward systems harder than normal reality, so ordinary life starts to feel flat.
The key point: the issue isn’t just “too much dopamine,” but that our baseline expectation for stimulation gets pushed so high that real life can’t compete.
TLDR:
If you want life to feel more real again:
- reduce engineered stimulation
- lower your baseline
- spend more time in lower-stimulation states
- stop judging real life against optimized digital experiences
https://x.com/thedarshakrana/status/2042643388642791893?s=46
Spoke to some friends who feel they are very left behind because of how fast things are. The reality is still we are in an echo chamber. The masses still have not caught on yet. We think we are lagging behind because of the algo we are being fed.
Modern life feels dull because our brains are getting trained by “superstimuli” - social media, hyper-edited content, instant novelty, vanity metrics, junk food, etc. These hit our reward systems harder than normal reality, so ordinary life starts to feel flat.
The key point: the issue isn’t just “too much dopamine,” but that our baseline expectation for stimulation gets pushed so high that real life can’t compete.
TLDR:
If you want life to feel more real again:
- reduce engineered stimulation
- lower your baseline
- spend more time in lower-stimulation states
- stop judging real life against optimized digital experiences
https://x.com/thedarshakrana/status/2042643388642791893?s=46
X (formerly Twitter)
Darshak Rana ⚡️ (@thedarshakrana) on X
How To Make Life Feel "Real" Again
China’s Ministry of Education has launched a comprehensive initiative to integrate AI throughout the national education system, spanning from K-12 to university levels.
Under this plan, AI will become a mandatory public course for college students, complemented by cross-disciplinary curricula at all grade levels.
The Ministry aims to establish a fully realized AI education framework by 2030.
Under this plan, AI will become a mandatory public course for college students, complemented by cross-disciplinary curricula at all grade levels.
The Ministry aims to establish a fully realized AI education framework by 2030.
❤2👍1