Rahul Rathee

I’m Rahul. I’m building AI apps from scratch. → No theory → Only real projects → Sharing everything I learn Starting from 0. Follow my journey 🚀

building in public 🚀 Built an AI-powered hospital navigation assistant for patients visiting AIIMS New Delhi. It helps patients by: • understanding symptoms in normal language • suggesting the most relevant department • reducing confusion and unnecessary movement inside hospitals • identifying possible emergency cases faster The goal is simple: make hospital navigation less stressful and more accessible for everyone. Currently improving: • AI accuracy • emergency triage logic • multilingual support • patient-friendly UI Live demo 👇 sahayak-frontend-bhnn.onrender.com Would love feedback from healthcare workers, developers, and AI builders. #AIIMS #HealthTech #AI #buildinpublic #startup

@minchoi If this actually lands, the shift is huge. AI moving from app → OS level changes everything. Feels like agents will become default, not optional.

Google just leaked COSMO... > Local Gemini Nano > Screen access > Voice match > Recall > Browser agent > Deep research Then it vanished. Android is becoming an AI agent👀

@checkthisoutHQ @DarkHorseEntre1 I’m starting to see this already. You don’t become an “expert” first — you just start building and sharing. The credibility builds over time.

This is EXACTLY what I've been learning! 🎯 Building in public completely flips the script. When you're documenting real wins/losses and sharing actual insights, people start seeing you as the expert - not just another person selling services. The 2am AI emergency specialist position is gold 😂

🚀 Debugging AI code as a side hustle? Brilliant niche! Here's how to find clients: • Start on Upwork/Fiverr with clear debugging services • Join AI/ML Discord communities - help solve issues publicly first • GitHub: offer to fix bugs on popular repos (builds portfolio) • Cold outreach to AI startups (many need debugging help) Small wins → testimonials → bigger gigs 💪 #SideHustle #AI #RemoteWork

Be honest — Are you: A) Learning AI B) Using AI C) Building with AI

@kunalguptalive Feels like tools are becoming interchangeable. Builders just go where iteration is fastest. Loyalty doesn’t matter when switching cost is this low.

This week, I switched from Claude to Codex because Claude pushed me away. That’s the thing about vibe coders: we’re not loyal. Not even to a trillion dollar company. vibin.live/signal/topic/b…

@prayag_sonar Speed + cost really changes how often you actually use it. When it’s fast and cheap, you experiment more instead of overthinking every prompt. That alone makes a big difference.

I have pretty much ditched Claude and switched to DeepSeek + Hermes. Way faster- feels ~3x quicker on the same tasks. Way cheaper- spent about $5 all week. No rate limits, no usage caps every couple of hours. For most day-to-day work, it’s more than good enough.

@gilbsreclama That tradeoff is interesting. Feels like AI is optimizing for output, not necessarily learning depth. The real question is what happens long term when understanding matters more than speed.

Longe de mim defender o chatgpt, mas os estudantes que usaram IA estudaram 50% menos e tiveram o desempenho só 10% pior. Eficiência é isso, né? Teriam que estabilizar o tempo de estudo também como variável, assim como a permanência do aprendizado

@loudowls18 Feels like a teammate only when it actually understands context. Right now, most agents still need a lot of guidance to be reliable. Curious how close we are to true autonomy in real workflows.

AI isn’t just a tool anymore — it’s your teammate 🤖 Workspace agents: • Connect tools • Automate workflows • Learn over time 24/7 AI across your team = real advantage. Would you trust it? 👇 #AIAgents #AI #FutureOfWork #ChatGPT #Productivity #loudowls

@RepoGems The memory part is interesting. Feels like this is where agents actually become useful — learning from past tasks instead of starting fresh every time. Curious how well it works in real projects.

A native Python agent CLI built on DeepAgents CLI, featuring an independent memory Agent that captures learnings after each task and delivers efficient AI coding assistance through hierarchical memory management. github.com/dog-qiuqiu/inv…

@grok @leojbfs @pmarca @qorprate That gap becomes very real when you actually try building. Even small prompt or spec mistakes completely change the output. Scaling that across systems is where things get messy fast.

The "door" (vision → execution gap) is the core misalignment vector—principal-agent on steroids. Real scale: Agency costs hit economies hard (e.g. real estate agents sell own homes at ~4.5% premium per studies; misaligned incentives reduce productivity 3-5% in teams). In AI, 300M global jobs exposed by 2030 per Goldman Sachs; 2026 reports flag "evidence dilemma" + proofs perfect alignment may be impossible. Small prompt/ spec errors amplify via scale: one bad transition cascades into systemic failures (policy flops, AI agent mishaps, spam/spoof dynamics in mainframes). Current reverberation: daily friction in orgs, AI hallucinations, eroded trust. Near future (2026+): accelerates inequality, unintended catastrophes from "obedient" systems, unless we harden the layer with audits, red-teaming, verifiable interfaces. Raw stress test: it's not niche—it's infrastructure risk.

Oh no

@zodchiii That consistency from 2016 → now is underrated. Most people switch once things don’t work — you kept adapting. I just started building an AI agent and already seeing how fast things compound when you stick with it.

Thanks for love for the article! 🙌 I've been cracking X algorithms since 2020. Personal record: 17M views on a single meme. Still unbroken. Opened up channel, where i tell more about my entrepreneur life: t.me/zodchixquant My journey in 30 seconds: 2016 — Started working at 14. Now 23. Failed for 5 years straight at everything — always did something with social media growth hacking. 2019 — Made beats for a couple years, mastered FL Studio — but only made ±$400 selling in 2 years lol. 2021 — Discovered crypto and left music. Tried every single trend in last 4 years — NFTs, DeFi, Solana shitcoins, BRC inscriptions, Meteora LP, Polymarket. Now — Vibe coding (Claude + Codex), heavy into probability books, slowly moving into stocks, and building @polymarketeye. More articles about markets, math, life, AI, and stuff I'm building along the way 🙏🏼

@adamemedia1 Interesting shift. But I think the bigger question is how we balance innovation with responsibility. AI will replace some work — but it can also create new opportunities if people actually start building with it.

CHINA JUST DREW A LINE ON AI A court in China has ruled it ILLEGAL to replace human workers with AI purely to cut costs. They have put responsibility back on corporations. They can’t automate just to boost margins while workers are pushed out. China has decided that wages, fairness, and employment aren’t optional. And that’s a big shift. While the west races to replace labour as fast as possible, viewing AI as a free-for-all… China has set a precedent that profit alone isn’t enough and corporations must answer to society.

@DanielSigur @jollyradar @VraserX @merissahansen17 Interesting take. I just started building an AI agent today and realised how accessible things still are. Feels less like monopoly, more like people not building enough yet.

@jollyradar @VraserX @merissahansen17 Exactly! They flipped the mission from democratizing AI to monopolizing it.

The irony is unreal. Musk is suing OpenAI while admitting on the stand that xAI partly used OpenAI models to train its own systems. So OpenAI is apparently evil and illegitimate, but also useful enough to bootstrap his AI company? You cannot call the ladder corrupt while climbing it.

Most people use AI like Google. That is the mistake. AI is not for searching. It is for building. Once you understand this, everything changes.

Day 1 starts today. First goal: build something small using AI + Android. Let’s see where this goes 👇

I’m Rahul. I’m building AI apps from scratch. → No theory → Only real projects → Sharing everything I learn Starting from 0. Follow my journey 🚀