prajwal

ex-early team @get_aspora (sequoia, greylock)

24, stablecoin infra, ai agents

[prajwal @PPaloju]

@lukaivicev @usesugar hey luka, sent you a DM!

[prajwal @PPaloju]

@lukaivicev curious on your thoughts on why wasn’t it possible to solve for it till now? why was it untapped?

[prajwal @PPaloju]

instinct is doing to ai agents what @meetgranola did to meeting notes

granola lowkey crossed the line by recording audio locally without asking for consent, but the convenience unlock was what made it *extremely sticky*

instinct is doing the same thing: browser access, logging into sites and auto-reading otps while poke and other ai agents play it safe and stop

[prajwal @PPaloju]

not chatgpt summoning euler to help me write cold emails for my b2b saas

[prajwal @PPaloju]

just can’t stop watching these robot olympic videos!

the future is here and china is a different place man

[Meer | AI Tools & News @Meer_AIIT]

Robot Olympics 💀

[prajwal @PPaloju]

read this profile of @chameleon_jeff back in april and been long hyperliquid:native ever since

https://t.co/iUNFuItfSP

[prajwal @PPaloju]

banger

[Shaurya Sinha @shaurya_sinha7]

This Independence Day 🇮🇳, I asked, "What will the next century of young Indians look like?"

$0.01 Drone Deliveries. UAVs that never land. Groceries in under 10 minutes.

Meet The 22nd Century Indian. A Documentary on a New India.

[prajwal @PPaloju]

students are cooked if this goes mainstream lol

[prajwal @PPaloju]

@pangram is shockingly good

[prajwal @PPaloju]

was trying @cursor_ai’s india-only starter plan, and the usage limits are… not great

the plan doesn’t quite make sense for a lot of users

with chatgpt luna pricing becoming so low, openai could launch a chatgpt go equivalent for codex, with an india-only plan with much better limits

@thsottiaux

[prajwal @PPaloju]

@vivekteega @WillowVoiceAI the reason to use willow was that it polished the text after transcription, which spokenly wasn’t very good at

[prajwal @PPaloju]

paying for speech to text doesn’t make sense anymore

was paying for wispr flow, then moved to spokenly (runs locally, but doesn’t polish text)

now on @WillowVoiceAI’s free plan. unlimited transcriptions that work well, and i don’t have a reason to upgrade

thin defensibility is probably why wispr added note-taking

credit where it’s due to wispr flow, they were pushing voice long before people realized talking, than typing would be the go to way to work with LLMs

[prajwal @PPaloju]

@ashutosh01113 🐅🐅

[prajwal @PPaloju]

@zarathustra5150 they’ve missed adding https://t.co/Kfe3NLZFxE haha

chatgpt never misses to search through papers from arxiv before giving an answer

[prajwal @PPaloju]

@Alex_Reinicke_ amazing! london needs more of these

almost all of them shut by like 7 pm

[prajwal @PPaloju]

@0xtrojan_ congrats guys! 🚀

[prajwal @PPaloju]

just realised the author of this in on twitter! @ijaack94

[prajwal @PPaloju]

chatgpt found me this really good notes of @rabois online

https://t.co/OmGK2iiDFB

[prajwal @PPaloju]

i was playing around with @Jackandjillai sometime last year

it’s a job platform where, instead of uploading a resume, you talk through your work

i was thinking about it today and realised i had told jack far more about myself than i ever would have told a recruiter or written on linkedIn.

voice is the clever part. you stop trying to package everything perfectly and just talk, so all the context that never makes it onto a resume comes out

jack & jill could end up knowing far more about the candidate than any job portal ever will - high quality data moat

super smart product

[prajwal @PPaloju]

@shreyashk_01 claude really needs a smaller / cheaper model to compete with luna - maybe the next version of haiku does it!

[prajwal @PPaloju]

gpt 5.6 luna high has been my daily driver, super smart model and very generous with its usage limits - 80% cheaper is nuts!

completely moved away from claude code to codex (+ openai removing the 5hr limits is a game changer)

[Sam Altman @sama]

major price cuts today:

*80% drop for GPT-5.6 Luna, now $0.20 per million input tokens and $1.20 per million output
*20% drop for GPT-5.6 Terra, to $2/$12
*GPT-5.6 Sol gets Fast mode in the API, up to 2.5x the speed for 2x the price, same intelligence