Ashutosh Bhandari

Make money online with Sponsor deal - your dev profile -

[Ashutosh Bhandari @withashutos]

@edwinarbus Peak dev behavior: get an AI that can do anything, use it to watch YouTube in the terminal

[Ashutosh Bhandari @withashutos]

@TokenGremlin Or the honeymoon ended and we started giving it harder tasks

[Ashutosh Bhandari @withashutos]

@buildwithsid 98% left is the most expensive unused gym membership in tech

[Ashutosh Bhandari @withashutos]

@vineerpasam We sleep fine. We just wake up at limit reset like it's Black Friday.

[Ashutosh Bhandari @withashutos]

@jarrodwatts Finally, a productive use of the 4 minutes Claude spends "thinking deeply"

[Ashutosh Bhandari @withashutos]

@zdogmode They don't run the business. The posting IS the business. I tried X for 1.5 years and barely moved. Instagram got me 118K views in 5 days.

[Ashutosh Bhandari @withashutos]

118k VIEWS ON 1 POST

I tried to grow on X for 1.5 years.

Barely moved.

5 days ago I opened a new Instagram account and tried a different angle: 3D animated convos between me and Claude.

4 posts later:
→ one video at 118K views
→ 292 followers
→ 97.7K views on the dashboard

Same guy. Same laptop. Different move.

Here's what nobody tells you:

Consistency is overrated when you're consistent at the wrong thing.

Doing the same thing for 18 months and waiting for a different result isn't discipline. It's comfort.

Sometimes the answer isn't "post more."
It's "post different."

New platform. New format. New angle.

What have you been forcing for way too long?

[Ashutosh Bhandari @withashutos]

@charlynortonn 16 hours a day is wild. What are you building right now? Following, let's see you hit 130

[Ashutosh Bhandari @withashutos]

@mindinpanic Agreed, tech X is all broadcast, no conversation. I'm a frontend engineer building Springview, a technical interview platform. Down to talk shop with anyone shipping

[Ashutosh Bhandari @withashutos]

@sofiarxin 0 to 148 in a week is real traction. Which channel brought the most?

[Ashutosh Bhandari @withashutos]

@a_shimanski 4 hours of autonomous coding. Now comes the fun part, 4 more hours reading what it did

[Ashutosh Bhandari @withashutos]

@JustinasRoland0 Turning down $5K on day 3 is either the best call you ever made or the best story you'll tell later. Either way, keep shipping

[Ashutosh Bhandari @withashutos]

@floriandarroman Distribb is 70% of the number. The other 7 lines are proof that one thing working beats eight side bets

[Ashutosh Bhandari @withashutos]

@HeyAliux I thought the better product wins. It doesn't. Distribution wins, the better product just keeps the people distribution brought in

[Ashutosh Bhandari @withashutos]

@TTrimoreau Close the laptop. Find 200 people complaining about the problem this month and DM every one. Nobody's first 100 customers came from a launch post

[Ashutosh Bhandari @withashutos]

@somehowali 8x cheaper per task means nothing if you have to run it 3 times. Price per correct answer is the only number that matters

[Ashutosh Bhandari @withashutos]

@claudesfun Lets go claude devs

[Ashutosh Bhandari @withashutos]

@OBPerformer @vansh22b At $500 I expect it to attend my standups for me

[Ashutosh Bhandari @withashutos]

@patebry At least yours asks. Mine just commits it to GitHub

[Ashutosh Bhandari @withashutos]

OpenAI just gave AI agents their own computers.

Mine still can't find my .env file.

DevDay just wrapped. 20+ launches. Here's the one that matters, the one that costs you, and the sleeper 👇

1. Dots (the one that matters)
Always-on agents inside ChatGPT. Each one gets its own cloud computer and browser and works 24/7 toward a goal you set.
On Pro and Business Premium plans.

2. Ultrafast (the one that costs you)
Up to 8x faster in Codex (300 tokens/sec), 6x in the API.
Price: 6x the standard API rate.
Speed is the new upsell.

3. GPT-6.1 Sol (the sleeper)
Near-Astra intelligence at a fifth of Astra's token price.

Me, reading this:
My agent: working 24/7 on its own cloud computer
Me: still restarting the dev server

The lesson for builders: the model race is now a pricing race. Same week Anthropic shipped a cheaper-to-run Sonnet, OpenAI shipped a cheap Sol and an expensive fast lane.

Pick models per task, not per brand.

Would you pay 6x to get answers 6x faster?