Twitxy Saved

Jaynit Makwana

Verified account This account is verified. Learn more

@JaynitMakwana

Sharing insights on AI, Productized Biz, Creator Economy, & No-code • DM for Brand Growth & Collabs | 📩 jaynit@scale10x.io
DM for collaboration 📩 waketheai.com/subscribe Joined March 2024
194 Following 60K Followers 31K Posts

Latest posts 5

Jaynit Makwana Verified account @JaynitMakwana
Four fashion visuals, and each one feels like it has its own creative direction.

MiniMax H3 Image is getting really good at this kind of editorial work. The styling, close-up details, and overall campaign aesthetic look surprisingly polished.

What's even better? 4K generation takes around 10 seconds, and it's 50% off right now.

I'd be making a lot more campaign concepts at that speed. @MiniMax_AI

#H3Image
2 replies 7 reposts 48 likes 26K views View on X
Jaynit Makwana Verified account @JaynitMakwana
a lot of data centers can't get any more power from the grid, so they can't add the hardware needed to serve bigger AI models.

Parasail just became the first inference cloud to bring Positron Titan online at scale.

Titan runs multi-trillion-parameter models on regular low-power memory, and it's air-cooled, so it fits the sites that can't take more power-hungry hardware.

I think power is going to decide who can serve the biggest models over the next few years.

#parasailpartner
Mike Henry Verified account @mikehenry8484 20h
Positron's Titan is joining the Parasail fleet. 5MW. Trillions of tokens a day of cost-optimized compute by late 2027. @parasail_io @positron_ai

@parasail_io is the first multi-silicon inference cloud that routes your prompts to the best chip for the job, serving trillions of tokens a day to hundreds of customers. Positron fills a vital part of compute spectrum on the cost optimized side: 4X better cost-per token than next gen NVIDIA/AMD systems which is perfect for large-scale agent swarms, background agents, and physical AI.

Positron has done incredible work building a chip and system that orchestrates across DRAM and SRAM allowing for massively scalable memory pools with commodity memory giving the best of both worlds for memory capacity and bandwidth.

We've been excited to partner with them in the early days through the launch of their Atlas chip and now big capacity plans with Titan.
@positron_ai
View quoted post on X
14 replies 24 reposts 150 likes 34K views View on X
Jaynit Makwana Verified account @JaynitMakwana
coding agents are great right up until the code looks correct and quietly breaks something important 😳

HUD is basically trying to give AI-generated code a much stronger second set of eyes before it ships.

very useful direction.
Roee Adler Verified account @roee 23h
“It looked good to me.” The five most dangerous words in the age of AI-generated code.

Just because it looks right doesn’t mean it’s ready to ship.

Who would’ve thought we’d make a musical about it at @hud_hq?
View quoted post on X
3 replies 3 reposts 52 likes 20K views View on X
Jaynit Makwana Verified account @JaynitMakwana
1 AI app shouldn't need a host, a database, a login provider, a job queue and a separate sandbox, each with its own bill.

Volcano puts all of it in 1 project.

you ask Claude Code, Cursor or Codex for something like a support agent that hands off to a person when it's unsure, and it builds it on Volcano and gives you a live link.
Marco Palladino Verified account @subnetmarco 1d
At Kong we run half a quadrillion AI tokens a year, and I kept seeing agents stuck on clouds built for websites.

So we built a cloud for agents.

Volcano is live today: the self-healing agent cloud. Sandboxes, durable workflows, Postgres, hosting.
View quoted post on X