Arman

photo

Antigravity pricing model is exceptional !

Other ones let you use ai until your monthly budget runs out.
Antigravity gives you a daily limit.

This means every day you can do a good amount of work with the AI model you like.

This is a huge shift !
#buildinpublic

Is there any YouTuber here?

I want to run past you an idea and get your feedback.
#buildinpublic

photo

I spent $100 on Facebook ads and I REGRET IT 😔

🌐 2 Ads
💲 3$ / day budget
🗓️ Ran for almost 25 days
🤬 0 Sales

What would you do differently?
#buildinpublic

I don't care what it says in benchmarks, but

Gemini 3.1 is dumber than Opus 4.6

.. and don't get me started on GPT 5.3-codex !
#buildinpublic

photo

4. Enable "Dependency Graph"

5. Enable "Dependabot alerts"

6. Done.

photo

1. Open repo on Github

2. Go to "Security" tab

3. Click on "Enable Depandabot alerts"

ALL your github repos are vulnerable UNTIL you do this:

Github has Dependabot.
It scans your repo regularly, and emails you if any vulnerability found.

It's disabled by default. This is how to enable it 👇
#buildinpublic

photo

Antigravity is exceptional 😲

It opens a browser and actually sees the changes it's making.

That is huge enhancement in results.

... and MicroPoster.so does all of these things.

New solutions for success in 2026

Comment below to get my special offer discount code 👇

3/ No one needs basic analytics and stats anymore!

They want insights ..
They want strategy done for them ..
They want reduction in their cognitive load ..

2/ Everyone hates AI !
No one likes AI generated contents.
Yes, It was exciting in 2025 ... but people hate it.

New social media management tools should have AI Co-Writer to enhance as per user's commands.

1/ Users don't want to figure out everything by themselves !
They want things done for them.

New social schedulers should predict what time works the best for each and every post and platform.

It's 2026 and users don't need old solutions for social media management.

here are 3 new solutions users are looking for 👇
(and every day it's trending more)

do you think vibe-coded codes are more secure ?

I really hope GeminiApp (on X) improves the memory of Gemini. It doesn’t update automatically.

Every time i have to give ALL the info about myself
#buildinpublic

photo

2 weeks after removing that, I started getting traffic from BING again.

So be careful guys, do not disallow "_next" folder in robots txt file or your website looks broken to Bing crawler.

photo

After lots of investigations, this is what i found.

I had added "_next" folder to my robots txt file.
This meant, when BING opened my website, its crawler didn't load static js files from "_next" folder, therefore the website looked broken to it !

Turns out Bing is very sensitive if HTML on your site is valid.

Any tiny issues, it removes you from SERP.

So I validated and fixed ALL HTML issues and warnings.

But this didn't fix it ....

photo

early Jan i realised my app (MicroPoster) is not getting traffic from Bing anymore.

jumped onto BWM (Bing WebMaster) and realised i was banned ! wtf !

I was getting crawled still but not shown in SERP, why the ban!? whyyy !?

photo

🤬 Bing banned my website for no reason !!!

this is why it happened and how I finally fixed it 👇
#buildinpublic

Reddit is so cooked!

Moderators have too much power.
It is NOT a democracy there.

we need an alternative.
#buildinpublic

If you’re building AI agents (like me) right now, this distinction will save you weeks of confusion.

What’s one thing that confused you when you started?

Why this matters:

If you only build tools → your agent is chaotic
If you only build skills → your agent is blind

You need both.

Real example:

MCP provides:
→ search_customer tool
→ billing_lookup tool

A Skill does:
→ understand the request
→ call the right tools
→ combine results
→ respond clearly

One exposes power.
The other uses it.

Skills are the opposite.

A Skill is packaged behavior.

It tells the agent:
• how to approach a task
• which tools to use
• what steps to follow
• how to handle edge cases

It’s actual capability.

Think of MCP like USB for AI.

It’s a standard way for agents to plug into:
• databases
• APIs
• files
• external services

MCP doesn’t tell the agent what to do.

It just says:
“Here are the tools available, here’s how to use them safely.”

If you remember ONE thing:

👉 MCP = how an agent connects to tools
👉 Skills = what an agent knows about how to do a task

Infrastructure vs ability.

Most people building AI agents mix up MCP and Skills.

They’re not the same thing.
They solve completely different problems.

Here’s the simplest explanation you’ll read 🧵
#buildinpublic #ai

I don’t know what you all are saying!

Codex 5.3 is very very very dumb!

Compared to Opus 4.6.

Opus works like an experienced engineer.

Codex works like a freshly graduated senior engineer.
#buildinpublic

Thanks