Sun retweeted
list of 15 best robotics researchers on X:
1. @DrJimFan, Jim Fan
2. @chelseabfinn, Chelsea Finn
3. @svlevine, Sergey Levine
4. @pabbeel, Pieter Abbeel
5. @drfeifei, Fei-Fei Li
6. @rodneyabrooks, Rodney Brooks
7. @0x_meden, Martin
8. @yukez, Yuke Zhu
9. @LerrelPinto, Lerrel Pinto
10. @SongShuran, Shuran Song
11. @tonyzzhao, Tony Zhao
12. @YunzhuLiYZ, Yunzhu Li
13. @xiaolonw, Xiaolong Wang
14. @GuanyaShi, Guanya Shi
15. @RemiCadene, Remi Cadene
Sun retweeted
God is an INFINITE MONEY PRINTER for anyone
quite literally the easiest meta on the market rn. all you have to do is:
1. find an existing product that's popular (health wearable)
2. make sure it's got a solid supply chain (you can buy these rings for $5-10 off of Temu lmao)
3. brand it with a † or anything else religious related
4. promise it'll bring you closer to god
5. run AI ads
whoever does this - thank me when you make your fastest $100k
Focus on micro-niches that tap into people's beliefs, identities, and values.
This brand is pulling in $113K a month selling emotions, not rings.
Sun retweeted
10 free Claude Code skills every builder should install right now (most people don't know these exist):
1.) Playwright Skill:
Claude opens your app in a browser, clicks through it and tests it
github.com/lackeyjb/playwrig…
2.) iOS Simulator Skill:
Claude builds, runs and taps through your iOS app on its own
github.com/conorluddy/ios-si…
3.) n8n Skills:
Claude builds n8n automations that actually work the first time
github.com/czlonkowski/n8n-s…
4.) Jezweb Skills:
up-to-date Cloudflare, React and Tailwind v4 patterns, so no more outdated code
github.com/jezweb/claude-ski…
5.) Trail of Bits Skills:
security audits from a top security firm, run on your own code
github.com/trailofbits/skill…
6.) Jeffallan Skills:
67 specialist skills covering your whole stack
github.com/Jeffallan/claude-…
7.) wshobson Agents:
a plugin marketplace where you install only what you need
github.com/wshobson/agents
8.) UI/UX Pro Max:
makes Claude's UI look designed not generic
github.com/nextlevelbuilder/…
9.) Superpowers:
makes Claude plan, test and debug like a senior engineer
github.com/obra/superpowers
10.) Skill Creator (by Anthropic):
turn any workflow you repeat into your own skill
github.com/anthropics/skills…
Bookmark this. Which one are you installing first?
Sun retweeted
stop asking GrokBot questions
start giving it work
I mapped 50 things you can actually delegate today:
→ research
→ content
→ operations
→ analysis
→ communication
→ coding
→ monitoring
→ support
→ planning
→ creative
plus the prompt formula, automations, best practices and where human approval still matters
50 tasks
10 categories
one sheet
save this before your next GrokBot task ↓
Sun retweeted
If you’re using Grok Bot, Dots, or Muse, you’re getting a fraction of what they can do.
They earn their keep when you give them work you already know well, WITH material you already have.
For example:
Grok Bot (content): “Read this approved transcript and compare it with our existing articles. Find a question our buyers care about that we haven’t answered well.”
Dots (product): “Walk through this as our target customer. Show me where the product stops helping them finish the job.”
Muse (paid): “Using these approved clips, this offer, and our customer description, recommend a clip for an ad.”
Once the output is useful, save the instructions and the corrections that got you there. Then hand it to someone on your team.
If I’m the only one who can use a bot, I’ve just made myself faster.
When someone else on the team can use it too?
We can change how the department gets work done.
Sun retweeted
If you have a file called context.txt that you paste into Claude every morning, read this.
You already built a memory system.
A manual one.
That you update by hand.
And forget to update.
Let Opus 5.5 do it.
It writes what it learns.
Cleans it up.
Loads only what each task needs.
Your context.txt, but automatic.
How to build it
Sun retweeted
Why is everyone with "building in public" in their bio completely broke?
Because 99% of people calling themselves "AI founders" are playing startup instead of running a business.
Locking themselves in a bedroom for 6 months. Write code, stress over Stripe webhooks, debug auth flows, and fight to get approved on the App Store.
All to launch a $49 a month SaaS tool to an audience of 0.
Building for ghosts.
Do the math on that business model:
To make $5,000 a month, you need 100 paying customers at $49 a month with zero churn. Every single month you are fighting tooth and nail just to replace the people who cancelled.
Meanwhile, look at how the real world operates.
I was sitting with 2 business owners in LA last week. They run a portfolio doing $15M to $20M a year. Real companies. Real payroll.
Their current internal "AI strategy"?
An intern using ChatGPT to draft reel scripts, and executives using it to rewrite emails.
They know they are falling behind. They know their operations are bleeding time. And they would cut a $5,000 to $10,000 check tomorrow to anyone who walks in, maps their actual bottlenecks, and automates their workflows.
A $5,000 check does not even require an executive sign-off for them. It is pocket change.
Stop playing SF startup founder for an empty room.
The playbook is simple:
1. Pick ONE client acquisition channel.
2. Build a working demo of a workflow that solves a real business bottleneck.
3. Show it to owners of businesses doing $5M to $20M.
4. Close the build and collect cash.
The market does not need another $49 wrapper.
It needs people who can actually IMPLEMENT.
Sun retweeted
Stop collecting AI resources.
Start using them. 📌
You probably have dozens of AI tutorials saved already.
But when it is time to actually learn, where do you begin?
Here is a simpler path.
10 GitHub repositories that can help you move from Python basics to machine learning, LLMs, AI agents, and practical AI projects.
1. Python-100-Days
Need to strengthen your Python?
Start here and build your fundamentals step by step.
github.com/jackfrued/Python-…
2. Generative AI for Beginners
A practical starting point for learning modern Generative AI.
You can explore LLMs, RAG, prompting, agents, and more.
github.com/microsoft/generat…
3. LLMs From Scratch
Want to understand how LLMs work under the hood?
This repository helps you learn the concepts by building them step by step.
github.com/rasbt/LLMs-from-s…
4. ML-For-Beginners
Before going deeper into AI, build your machine learning foundation.
This is a good place to start.
github.com/microsoft/ML-For-…
5. OpenAI Cookbook
Ready to move from learning to building?
Explore practical examples and guides for creating AI applications.
github.com/openai/openai-coo…
6. Stable Diffusion
Interested in AI-generated images?
This repository is worth exploring if you want to understand image generation models.
github.com/CompVis/stable-di…
7. AI Agents for Beginners
AI agents are becoming an important part of modern applications.
Learn about tools, RAG, agent frameworks, and multi-agent systems.
github.com/microsoft/ai-agen…
8. AI-For-Beginners
Want a broader understanding of AI?
Explore topics including neural networks, NLP, computer vision, deep learning, and classical AI.
github.com/microsoft/AI-For-…
9. LLM App
Want to build practical LLM applications?
Explore RAG, vector search, AI pipelines, enterprise search, and real-time data.
github.com/pathwaycom/llm-ap…
10. Segment Anything
Interested in computer vision?
Explore image segmentation and how it can be used in modern AI applications.
github.com/facebookresearch/…
A simple learning path:
→ Starting from zero → Python-100-Days
→ Learning ML → ML-For-Beginners
→ Understanding AI → AI-For-Beginners
→ Exploring LLMs → LLMs From Scratch
→ Learning Generative AI → Generative AI for Beginners
→ Exploring agents → AI Agents for Beginners
→ Building applications → OpenAI Cookbook + LLM App
→ Exploring computer vision → Segment Anything
You do not need to learn everything at once.
Pick one repository.
Spend time with it.
Build something small.
Make mistakes.
Try again.
Then move to the next one.
The goal is not to collect resources.
The goal is to actually build.
Save this list. 📌
It might come in handy the next time you sit down to learn AI.
Sun retweeted
INSTEAD OF WATCHING NETFLIX TONIGHT.
Spend 1 hour with this.
Claude AI FULL COURSE that teaches you how to BUILD and AUTOMATE anything.
The people who watch this tonight will wake up tomorrow with a new skill. Watch it and bookmark it now
Sun retweeted
Anthropic engineers just released the best 1-hour course on building agents with Opus 5.5:
"Opus 5.5 is the best model for building agentic systems."
00:00 - intro to Managed Agents
07:45 - Opus 5.5 brain + architecture
23:40 - skills, loops, memory graphs
54:00 - agent dreaming and orchestration
Prompts → Agents → Loops → Graphs
most people will stop after the architecture chapter and call it an agent
the last twenty minutes are where the agents start running each other
same model, same tokens, completely different week
watch it today
the full article is below, save it while it is still early ↓
Sun retweeted
if I had to make my first $1,000 online again in 2026, I wouldn't build a startup
fyi: in August alone, I paid $37k to affiliates and creators who market my products with their content
one of my users has even made over $75k just by setting up my $99/month product
so here are the 5 things I'd do instead:
1. set up AI agents for boring local businesses
a dentist doesn't want "an AI agent". he wants his phone answered, his reminders sent and his reviews replied to
I run 8 agents across my own companies, and most local businesses still run zero
At Squad, we have partners getting paid $3k every time they set up Squad for a client 🤯 (dm me if you want to do this btw)
knowing 20 plumbers is what actually takes work
2. remake ads that already work, for businesses that can't make their own
search your client's competitors in Revid's Meta and TikTok ad viral search and save only the ads that have been running for 30+ days
nobody keeps paying for an ad that isn't working, so those are the proven winners
remix those ads with Revid for your client, and they get to skip months of testing
3. get paid per view, not per follower
on Clipatra, brands pay creators for every verified view. in the first 3 weeks of beta, creators posted 38 videos, got 500k views and earned $4,266, which works out to roughly $8.50 per 1,000 views
you don't need a big account for this, you just need one video that works
4. promote the tools you already pay for
most of that $37k I paid in August didn't go to professional marketers. it went to people who already used my products every day and just said so in public
if you're paying for a tool you love, you're probably sitting on an affiliate link you haven't claimed yet
5. start a blog on one boring niche and let it pay rent
pick a topic you’ve actually lived. it doesn’t need to be ambitious, it just needs to be something you understand better than almost everyone searching for it.
it can be as simple as coffee. if you know coffee better than 99% of the world does, you can start a blog about it
once the blog hits DR 30+ and 100+ clicks a month for 3 months straight, you can plug it into Outrank
it finds the advertiser, writes a full article in your tone with one link inside and publishes it on your blog
you get $1 per DR point per article, paid straight to your bank through Stripe
so a DR 60 blog makes $60 an article. at 3 a day, that's up to $5,400 a month without you writing a single word
Sun retweeted
Paste this prompt into Claude Opus 5.5.
It turns Claude into an employee that knows your rules, your voice and your work, and runs tasks while your laptop is closed.
You’re welcome:
Sun retweeted
1 model + 7 new GitHub repos = a completely different AI stack
GPT 6 Astra can reason
these repos give it routing memory browsing validation and visibility
01 LiteLLM
▸ github.com/BerriAI/litellm
→ model gateway + routing + fallbacks
AGENT CORE
02 Mastra
▸ github.com/mastra-ai/mastra
→ agents + workflows + tool coordination
MEMORY
03 Letta
▸ github.com/letta-ai/letta-co…
→ stateful agents with persistent identity + memory
BROWSER ACTIONS
04 Browser Use
▸ github.com/browser-use/brows…
→ agents that navigate and operate real websites
SAFETY
05 Guardrails
▸ github.com/guardrails-ai/gua…
→ validate inputs outputs + trigger corrective actions
OBSERVABILITY
06 Phoenix
▸ github.com/Arize-ai/phoenix
→ traces + experiments + evaluations + debugging
DEVELOPER SURFACE
07 Continue
▸ github.com/continuedev/conti…
→ run custom coding agents across IDE terminal + CI
the architecture:
request → model gateway → agent core → memory → browser actions → validation → trace → evaluate → improve
the interesting part is no longer choosing one model
it is choosing what surrounds the model when real work begins
Sun retweeted
THIS IS F*CKING INSANE, I CASHED OUT $60,000 AND SLEPT, MY GROK BOTS SOLD A TRUMP COIN 90 MINUTES BEFORE IT CRASHED 78%, $1,873 IS NOW $8,977
in 3 nights Jev + my Grok bots turned $100 into $61,873
the replies said nobody ever cashes out, it's just numbers on a screen
so yesterday i moved $60,000 to my bank and left them the change, $1,873
if they're actually good, the change is enough
the coin:
VeilTrump, a Trump meme coin, 1 day old, $866K at 1pm
the night:
→ 1:35pm first trade made $45, i almost turned them off
→ 3:20pm the dip to $714K, bought, sold $1.71M an hour later, +$2,400
→ 7:05pm $1.32M to $2.10M, +$2,615
→ 10:30pm skipped a coin up 218% in 40 minutes, its pool got emptied at 11:30pm
→ 12:25am $1.27M to $1.42M, +$777
→ 5:00am sold at $1.10M, then all cash
the morning:
→ 6:30am the coin fell from $1.03M to $230K in 25 minutes
→ my bots had been out for 90 minutes
$1,000 held in that coin is $265
my $1,873 is $8,977, 4.8x on a coin that lost 73%
honestly i laughed at the 5am exit until i saw 6:30am
at 0:09 my spider locks the $60,000 in a silk safe
at 0:24 the coin falls off the screen, my bots already in cash
$100 → $61,873 → $60,000 to my bank → $8,977 still working
every prompt and the full bot setup are in the article i quoted below ↓ free
they trade on one exchange, it's my first reply, sign up there and the bonus is yours
bookmark this, one morning your coin drops 78% before you're up
send it to the friend who never sells
would you have cashed out the $60,000 or let it ride?
Sun retweeted
10 GITHUB REPOS FOR TURNING THE PUBLIC WEB INTO USEFUL DATA.
1. Firecrawl — turn websites into AI-ready data.
Crawl web pages and convert content into clean Markdown for search and AI workflows.
github.com/firecrawl/firecra…
2. Oxylabs Web API MCP — search, extract and structure live web data.
Connect AI agents to web search and scraping tools through MCP. Particularly relevant for market research, competitive intelligence and automated web research workflows.
github.com/oxylabs/web-api-m…
3. ScrapeGraphAI — describe the data you need.
Use LLM-powered workflows to extract information from websites.
github.com/ScrapeGraphAI/Scr…
4. Crawl4AI — crawling designed for AI pipelines.
Convert web pages into structured, LLM-friendly content.
github.com/unclecode/crawl4a…
5. Newspaper4k — extract news articles automatically.
Retrieve article text, metadata and useful information from news URLs.
github.com/AndyTheFactory/ne…
6. Beautiful Soup — make HTML data easier to work with.
Parse web pages and extract specific elements with Python.
github.com/wention/Beautiful…
7. Extruct — extract structured data from websites.
Parse embedded metadata formats such as JSON-LD, Microdata and Open Graph.
github.com/scrapinghub/extru…
8. Pricefy — explore product price monitoring.
Investigate tools and workflows for tracking prices across online stores.
github.com/pricefy/pricefy
9. Pandas — turn collected data into insights.
Clean, transform and analyse tabular datasets with Python.
github.com/pandas-dev/pandas
10. DuckDB — analyse data without a heavyweight database.
Query large datasets locally using an efficient analytical SQL engine.
github.com/duckdb/duckdb
All are open-source.
The real advantage isn't collecting more data. It's turning it into better decisions.
Sun retweeted
Don't waste 2 years learning to become an AI agentic engineer in 2026.
Andrew Ng, the godfather of AI, gave the complete playbook to become one from scratch.
1 hour course. Free:
• 00:00 - AI agent basics
• 12:12 - AI Agentic workflows & design patterns
• 53:27 - Practical tips for building AI agents
• 1:20:30 - self-improving AI agent loops
• 1:30:19 - multi-agent AI systems
I watched it last night.
Halfway through, I realized I could get into Anthropic in weeks, not years.
Bookmark now. Watch it. Then build your own AI agent
Sun retweeted
Be an AI Engineer in 2026. Learn this:
1. The Vector Stack:
- Embeddings (Text-to-Vector)
- Vector Databases (Storage)
- Semantic vs. Hybrid Search
2. RAG & Context:
- Advanced Retrieval (Reranking)
- GraphRAG (Knowledge Graphs + Vectors)
- Long-context window management
3. Pipelines & Orchestration:
- Chaining (Deterministic flows)
- Routing (Selecting the right model for the task)
- Frameworks: LangChain / LlamaIndex
4. The Agentic Layer:
- Tool Use (Search, APIs, Code Interpreters)
- Planning Loops (Reasoning before acting)
- Multi-Agent Orchestration (Swarms)
5. Evaluation & Testing:
- Golden Datasets (Establishing Ground Truth)
- LLM-as-a-Judge (Using strong models to grade weak ones)
- Metrics: Faithfulness, Answer Relevance, Recall
- Continuous Eval in CI/CD pipelines
6. Ops & Monitoring:
- Tracing (Debugging the chain)
- Cost & Latency optimization
Want help? Then attend this:
🚨 Want to learn how to build + ship AI and Data Science projects (that businesses actually want in 2026)?
On October 21st, I am hosting a free workshop to help you get started with AI + DS projects in Python (free).
Register here (500 seats): learn.business-science.io/ai…
