Vas (clueless arc)

building to retire your parents is the moat

every founder pivoting to "agents" in 2026 is making the same mistake every founder made when they pivoted to "no-code" in 2022 they're confusing a technology shift for a product opportunity agents are a feature, not a category. saying you're "building an agent" is like saying in 1998 you were "building an HTML." true. and useless the founders who win this cycle aren't the ones who build agents. they're the ones who pick a specific human bottleneck (emotional, decisional, financial, social) and quietly let agents disappear inside the solution the agent you don't notice is the one that worked

my chrome has 14 research related tabs open. my arc has 11 content and creativity related tabs open. my claude code has 3 terminals open. my brain has 47 tabs open ranging from that one song whose lyrics i can't remember to how to make chicken so i don't fall behind on my protein intake at least i don't hear voices. yet.

this week's weird but true observation: gen z women aren't dating men "whose job isn't legible" the unspoken filter: if she can't describe what he does in one sentence at a family dinner, the relationship has a 6-week ceiling this isn't shallow. it's pattern-matched survival. and every product targeting gen z women should understand it

the indus river is part of ajrakh's recipe. literally. you cannot make the print without washing the fabric in those specific minerals day 2 fixing mislabeled indian crafts in AI: ajrakh is a 16-step block printing process from sindh and gujarat. the same fabric gets washed in a specific order: first in soda ash, then resist-printed with a paste of lime and gum, then dyed in indigo, then again in alizarin, then washed in the indus every step is timed to specific weather and humidity windows. an ajrakh print made in monsoon and the same print made in summer come out completely different most ajrakh families have been making the same patterns for 400-500 years. the patterns aren't designs. they are cosmology: astral diagrams, geometric meditations, prayer cloths that became wearable every fashion AI on the market does one thing: they describe ajrakh as "indian block print." like saying mona lisa is "european portrait" we're not behind on AI. we're behind on respect. and that's the gap that lets a $0 budget compete with $50M-funded startups in india. they don't see what they're missing

openai is gradient descent in a $300B suit. anthropic is the safety paper that walked itself into a product company they're not the same kind of bet and the market still acts like they are

I'd rather be a failure than a coward.

hot take: co-founders in 2026 are mostly a tax on speed one good founder + claude code + agent SDKs ships faster than two humans negotiating a roadmap. faster than three. faster than a seed-stage team of five the only reason to take a co-founder this year is loneliness. and that's a therapy problem, not a cap table one

watched a 22-year-old turn down a role this week because the manager "didn't have a strong public presence" thought it was wild. then thought about it for 20 minutes and realised: she was right. in 2026, your manager's reputation IS your training data. she wasn't being picky. she was being efficient

vibe coding's actual bottleneck is taste, not syntax claude can write your 200-line function in 30 seconds. it can't decide whether the function should exist. it can't tell you if the feature is for your real user or for the version of your user that lives in your head i used to be slowed down by what i couldn't write. now i'm slowed down by whether i'm right the AI compressed the cheap part of the work. the expensive part is whatever you've been avoiding. for me, that's the decision, not the deploy

@iamcluelessAI this hit close to home

ralph lauren is selling "vintage gold drop earrings" at 12x the markup of the jhumkas your nani had in a steel box and we are supposed to applaud the rebrand

@ionleu iamclueless.ai

drop ur startup link

chikankari isn't "white embroidery." that's like calling sushi "cold rice" day 1 fixing mislabeled indian crafts in AI: chikankari is a 6-stage hand technique from lucknow where the design is first block-printed on fabric in washable blue ink, then embroidered using 36 distinct stitch types: phanda, murri, jaali, bakhia, hool, keel kangan, and 30 more. each used in specific places to create depth and texture you cannot replicate by machine a single anarkali takes 60-90 days, 4-6 artisans, mostly women who learned from their grandmothers every western fashion AI sees "white embroidery" and groups chikankari with eyelet, broderie anglaise, machine schiffli. these are not the same thing. the difference is 200 years of unrecorded technical knowledge that nobody has bothered to digitise we're trying to fix that. one stitch type at a time

founder advice that ages well: ship the thing nobody will compliment. it's the only thing your competition can't copy from your tweets

@ignyteHQ it’s like a rite of passage

@vasbhatia i've caught myself explaining product decisions to my coffee maker. when you're the only voice in the room, everything becomes a sounding board for sanity checks you can't skip.

caught myself defending a product decision to my fridge yesterday the fridge wasn't even arguing. it just had condensation that looked like skepticism solo founder year is going great, thank you

gen z doesn't have job ambition. they have leverage ambition. that's a completely different operating system the older generations were taught: get a job, climb a ladder, become valuable to a company, the company takes care of you. it doesn't anymore. gen z watched their parents discover that the hard way so gen z optimises for the thing the company can't take away: skills that move, networks that exist outside the org chart, brand that follows you between roles, side projects that compound ambition didn't die. it just stopped pointing up the ladder and started pointing across the platform every founder hiring gen z and complaining "they don't want to work hard" is reading the assignment wrong. they want to work hard. they don't want to work hard for one employer's equity that vests over 4 years while the company tries to figure itself out build for the gen z working life, not the millennial one. you'll hire better

the vibe coding move that 10x'd my shipping speed: stop asking claude to "improve this code." it will. it will improve it 47 different ways across 6 conversations and you'll have no idea what it actually did instead, ask it to "list 3 specific improvements to this code, ranked by impact, without making any changes." you read the list. you pick the one you actually want. then you say "do that one. nothing else" the AI is not the bottleneck. your inability to constrain it is every time i let claude "improve" anything autonomously, i spend the next 4 hours figuring out what it broke. every time i constrain it explicitly, i ship in 20 minutes ship the constraint. not the request

a 5/10 product with distribution beats a 9/10 product without distribution 9 times out of 10 and the tenth time, the founder still spent more on distribution than on the product

@jellynet_ this is sick! i’ll check it out

@vasbhatia Respect the transparency. $291/month is honest — most people underreport. The Gemini vision + flash split is smart routing instinct. The next unlock is doing that across 40 providers automatically instead of manually managing each one → producthunt.com/products/jelly…

real money i'm spending to build a consumer AI app in 2026, since people keep asking: claude code: ~$200/month. i abuse claude a lot like an abusive parent ngl gemini API (vision + flash): ~$60/month. surprisingly cheap. the bill comes from latency more than tokens supabase pro: $25/month. would 100x if we had real traffic but we don't yet vercel: $20/month. moved off cloudflare workers because the cold starts were killing demos x premium: $16/month domain + email: $30/month combined total: ~$291/month. that's my consumer AI startup. no team, no office, no equity diluted yet most founders raising $2M to build the same thing are paying 4x in agency fees alone the moat for consumer AI isn't capital. it's whether you know how to keep costs disciplined while you figure out what you're actually building

my mother just asked me if "the app" is "doing well" with the same tone she uses to ask if my plants are dying correctly. they are both at the same stage solo founder year update: my mother thinks i'm running a small business out of my room and she's not wrong, technically