Shubh Jain (@shubhdj191) 's Twitter Profile
Shubh Jain

@shubhdj191

Full Stack Developer | Innovating with MERN Stack, Next.js & AI | Building Scalable Solutions

ID: 1094057210756313088

calendar_today09-02-2019 02:15:40

253 Tweet

22 Takipçi

42 Takip Edilen

Shubh Jain (@shubhdj191) 's Twitter Profile Photo

AI Agents compared: PERPLEXITY COMPUTER → 19 models integrated → Cloud-based (managed risk) → $200/month → Turnkey solution → Enterprise-ready OPENCLAW → Locally installed → User manages risk → Free + open source → Full data control → Requires technical skill CURSOR

AI Agents compared:

PERPLEXITY COMPUTER
→ 19 models integrated
→ Cloud-based (managed risk)
→ $200/month
→ Turnkey solution
→ Enterprise-ready

OPENCLAW
→ Locally installed
→ User manages risk
→ Free + open source
→ Full data control
→ Requires technical skill

CURSOR
Shubh Jain (@shubhdj191) 's Twitter Profile Photo

I saw a video this week that made me seriously question everything about AI. And not in the AGI is coming for us way. More like… are we building tools or our own digital overlords? Google dropped Nano Banana 2. It makes images FAST. Like, twice as fast. and free. But the

Shubh Jain (@shubhdj191) 's Twitter Profile Photo

this Week of AI is Insane. - nano banana 2: image gen 2x faster & free - perplexity computer: ai agents automating whole projects rivaling bloomberg terminals - anthropic vs pentagon: fighting over ethical safeguards for ai weapons + surveillance. - model distillation:

Shubh Jain (@shubhdj191) 's Twitter Profile Photo

Google just dropped Nano Banana 2. And it's actually good. Key features: → 2x faster than original → FREE via Gemini → Search grounding (pulls from real-time web) → Better text rendering than Midjourney → Multiple style options Tested it this morning: Prompt: "Enamel

Shubh Jain (@shubhdj191) 's Twitter Profile Photo

Microsoft Co-Pilot Tasks just launched. (Early access only - waitlist) What it automates: → Document generation → Recurring research → Meeting scheduling → Multi-step workflows Sounds like every other AI tool? Here's the difference: It's ENTERPRISE. Which means: → IT

Shubh Jain (@shubhdj191) 's Twitter Profile Photo

Be delusionally optimistic about your potential. Be delusionally optimistic about your potential. Be delusionally optimistic about your potential. Be delusionally optimistic about your potential. Be delusionally optimistic about your potential. Be delusionally optimistic

Shubh Jain (@shubhdj191) 's Twitter Profile Photo

Stop using AI for the things that make you valuable. Most people are using AI to replace their thinking. That's a mistake. I use it to replace my labor. My simple rule for high-growth workflows: ⚡ Scale → AI (Data synthesis, summaries, repetition) ❤️ Trust → Human

Shubh Jain (@shubhdj191) 's Twitter Profile Photo

If AI makes your work 10x faster, and everyone else's work 10x faster, nobody's 10x ahead. The margin collapses. The new margin is judgment. Relationships. Vision. The things that don't scale through prompts. Build those while everyone else is racing on the same flat track.

Shubh Jain (@shubhdj191) 's Twitter Profile Photo

AI is a fantastic tool that saves us from tedious labour, but if you don't use the spare time for proper, human thought, you're just accelerating the inevitable slide into beige intellectual wallpaper.

Shubh Jain (@shubhdj191) 's Twitter Profile Photo

Beginner vibe coder: - installs claude code/antigravity/codex/cursor - thinks about what to build - prompts Claude - begin build - generates an AI slob and debugs endlessly Not so beginner vibe coder: - installs claude code/antigravity/codex/cursor - setup personal

Shubh Jain (@shubhdj191) 's Twitter Profile Photo

8 Python Libraries That Made My AI Agents Actually Work 1. LiteLLM - Because Provider Lock-In Is for Suckers 2. Instructor - Because LLMs Lie About JSON 3. Tenacity - Because APIs Will Betray You at 3am 4. Logfire - Because Processing Complete Is Not Debugging 5. Diskcache -

8 Python Libraries That Made My AI Agents Actually Work

1. LiteLLM - Because Provider Lock-In Is for Suckers
2. Instructor - Because LLMs Lie About JSON
3. Tenacity - Because APIs Will Betray You at 3am
4. Logfire - Because Processing Complete Is Not Debugging
5. Diskcache -
Shubh Jain (@shubhdj191) 's Twitter Profile Photo

that python bootcamp you took for ai? probably useless knowing how to write a for loop doesn't teach you how llms actually work most non-technical people in ai are still clueless about rag or context windows you don't need to code but you do need to speak the language