AY Automate

We help you build or hire the best AI Agent for your unique needs.

Our working process is so simple:

 1. In one hour, we analyze your business.
 2. We figure out your tech setup.
 3. Our AI agents check your workflows( manual ones) 
 4. We find weak spots and new opportunities.
 5. We send you a detailed report.

What happens next?

You decide what to improve.

Co-Founder | Chief Technology Officer

Humanoidz

Building autonomous ventures. Sharing what i will be learning

Co-Founder ( Human ) 

KohortPay

At KohortPay, I worked on some of the most exciting challenges of my career. 

But what truly made it special? The incredible team I worked with.

At first, the company was focused on payment solutions. But after a few months, we made a big shift—we decided to focus completely on referral programs t. I was deeply involved in making sure this change was a success.

➔ 𝗦𝗵𝗼𝗽𝗶𝗳𝘆 𝗔𝗽𝗽: Built a fully functional app in just 15 days. First time working with Shopify? No problem. Thanks, Cursor!

➔ 𝗕2𝗖 𝗮𝗽𝗽 𝗳𝗼𝗿 𝗰𝘂𝘀𝘁𝗼𝗺𝗲𝗿𝘀: Improved the app’s design and user experience using AI-driven analytics to identify pain points and optimize workflows, making it easy and enjoyable to use.

➔ 𝗘𝗺𝗮𝗶𝗹 𝘁𝗲𝗺𝗽𝗹𝗮𝘁𝗲𝘀: Created and Maintained 15 Default Email Templates for our brands.

➔ Brand dashboard: Added powerful features like customization of 𝗕2𝗖, 𝗶𝗻𝘃𝗼𝗶𝗰𝗶𝗻𝗴, 𝗳𝗶𝗹𝘁𝗲𝗿𝗶𝗻𝗴, 𝗮𝗻𝗱 𝘁𝗿𝗮𝗻𝘀𝗮𝗰𝘁𝗶𝗼𝗻𝗮𝗹 𝗲𝗺𝗮𝗶𝗹𝘀 𝗮𝗹𝗹 𝗻𝗮𝘁𝗶𝘃𝗲.

➔ SDKs & documentation: Developed tools (SDKs) to simplify developer integrations.

The most exciting project? 𝗞𝗼𝗵𝗼𝗿𝘁𝗥𝗲𝗳 / 𝗞𝗼𝗵𝗼𝗿𝘁𝗜𝗻𝗳. We built it in just 𝘁𝗵𝗿𝗲𝗲 𝗺𝗼𝗻𝘁𝗵𝘀. It quickly became a key selling point and helped us sign deals with major brands like 𝗠𝗮𝗿𝘀𝗲𝗶𝗹𝗹𝗲 and 𝗖𝗮𝗿𝗿𝗲𝗳𝗼𝘂𝗿.

Want to explore our solution? Check it out here: KohortPay.com

If I had to summarize my contributions, here’s what stands out:

✅ Worked on 10 𝗼𝘂𝘁 𝗼𝗳 𝘁𝗵𝗲 𝗰𝗼𝗺𝗽𝗮𝗻𝘆’𝘀 12 repositories (except PHP, which was managed by Aymeric).
✅ Built an AI chatbot for the B2B dashboard to assist brands with quick actions.
✅ Developed an AI writer for customers to help them craft referral messages that drive more referrals.
✅ Helped onboard 10+ 𝗺𝗮𝗷𝗼𝗿 𝗰𝗹𝗶𝗲𝗻𝘁𝘀, making their setup process faster and smoother.
✅ Delivered KohortRef in just 3 months, securing deals with brands like 𝗠𝗮𝗿𝘀𝗲𝗶𝗹𝗹𝗲 𝗮𝗻𝗱 𝗬𝘃𝗲𝘀 𝗥𝗼𝗰𝗵𝗲𝗿.

Full Stack Engineer

Bevinzey

When I joined Bevinzey, I was handed a challenge: build an AI-powered platform to transform how students learn and teachers teach. Dr. Michael Danquah had the vision, and in just three months, we turned it into reality.

Here’s what I worked on:

➔ Seven AI Tools: A chatbot for instant answers, a “chat with documents” feature (upload PDFs and interact with them), tools for summaries, content creation, and even voice cloning and image generation.
➔ Growth Features: Designed a referral program to drive signups and a CEO dashboard with real-time data on users, subscriptions, and referrals.
➔ Platform Optimization: Integrated Stripe for smooth subscription handling, boosted SEO for better visibility, and set up Clicky analytics to track engagement.

The result? We launched in record time, and within weeks, 4,000+ users were on board. Features like document chat and voice cloning put Bevinzey ahead of the curve, setting a new standard in edtech.

Bevinzey wasn’t just a project—it was a leap forward in learning. 

See for yourself: Bevinzey.com.

AI & Full Stack Engineer

SaniChat

Sanichat wasn’t supposed to be a big deal. It was just an idea I had while playing around with GPT-3’s API. But then I thought, “Why does this AI stuff feel so complicated for everyone who’s not a developer?”

And then it hit me: What if I could bring AI to something everyone already uses? Like WhatsApp. Simple. Familiar. No need to explain “APIs” to my parents or my friends.

Three weeks. That’s how long it took me to build it. Three weeks of coding, coffee, and convincing myself I wasn’t completely insane.

The result? Sanichat—a WhatsApp assistant that could:
• Chat with your documents (send a PDF, and it gives you instant answers).
• Talk back to you (voice-to-text? Yes. Text-to-voice? Absolutely).
• Help with job searches (career advice, recommendations, all through chat).
• Oh, and show non-tech folks that AI isn’t scary—it’s actually useful.

It wasn’t perfect, but it worked. I even made a landing page and ran some ads to see if people cared. Turns out, they did.

People started trying it. Sharing it. Asking me things like, “How does it know that?” (Spoiler: AI magic). I also registered the company in London because, well, if you’re gonna do it, you might as well do it properly.

But here’s the thing: Sanichat wasn’t just about what I built—it was about what it taught me.

I realized I could take a wild idea and turn it into something real in no time. I learned how to simplify complex things for people who just want results, not lectures. And most importantly? I discovered how much I love building tools that actually help people.

Sure, I paused the project when life got busier.

But I'll always remember it as the moment I stopped just "thinking big" and started building big.

Because sometimes, you don't need a perfect plan.

You just need: Idea ➛ 3 weeks ➛ Some coffee.

Then dedication to finish that ....

Founder ( Sold ) 

Binko (ex Mister Bin)

Building a Waste Management Revolution? Sounds Crazy. But We Did It.

An app that could transform how entire cities handle trash. That's Binko.

Working with Google's AI team, we created something wild: an intelligent waste management platform that does what others thought impossible.

Features that blow your mind:

➔  AI-powered product recognition (2 MILLION products mapped!)
➔  Real-time trash truck notifications
➔  Embedded city maps showing every single waste container
➔  Precise waste categorization based on product composition


The founder's vision? Transform recycling from a chore to a smart, seamless experience. My job? Turn that vision into reality.

Investors saw more than an app. They saw a revolution.

Numbers that show we're making a real difference:

- 2 Million+ products in our database
- 96% sorting accuracy ( AI Model ) 
Potential to cut 10,000 tons of CO2 annually
$500,000 raised for future waste tech

We're not saving the world. We're just making it a little bit easier to take care of it.

Better together. 🌍♻️

See some Screenshot of the app down 👇

CTO | Full Stack Engineer

Sponsio® 

Sponsio started as a simple idea.

After endless meetings with ESG experts, the vision became clear.

Companies spend $100K annually on ESG reports. But are they reliable?

Stakeholders needed transparency. Trust was missing. Blockchain offered the solution.

We built a platform connecting companies, stakeholders, nonprofits, and associations.

Everything was logged on the blockchain—for trust, transparency, and accountability.

The MVP architecture prioritized simplicity, user experience, and seamless interactions.

We developed a ReactJS front-end for intuitive user interfaces.

The Firebase-powered back-end ensured smooth operations with their amazing firebase SDK ( firestore, realtime database, auth...)

Smart contracts on Polygon secured data and enhanced transparency.

I deployed and tested everything—from frontend to backend to blockchain.

Each feature was designed to simplify ESG validation and engagement. ( Solidity ) 

This project taught me problem-solving, innovation, and entrepreneurial execution.

Sponsio turned ESG reporting into a transparent, stakeholder-driven process.

CTO | Web & Blockchain Engineer

BIGBANG ITS

Hey there! Let me tell you about my time at BIGBANG ITS.

As a hosting provider, we weren't just another tech company. Our digital infrastructure was impressive:
- Over 2,000 active clients
- 10,000 registered domain names
- Comprehensive cPanel hosting solutions
- Robust web services across .ae and international markets

Here's a fun story: We tried to build the "Shopify of the Middle East". Totally ambitious, right? We didn't quite nail it, but man, what a ride! One of our memorable clients during this adventure was Sedi Hisham - they were part of our wild experiment.

My role? I was the software engineering guy who made all this possible. I worked on websites like Argaam.com, Chomie.net, and Tiffany Flowers - each project solving real business challenges.

Our domain management was a game-changer. We weren't just selling hosting; we were providing complete digital ecosystems. Our cPanel solutions gave clients everything they needed - from domain registration to advanced hosting packages.

My days were about more than just coding. I was the translator between complex tech and business needs. Every week, I'd meet with clients, breaking down technical details into language everyone could understand.

APIs were my specialty. I built systems that made different software talk to each other seamlessly. Imagine creating perfect communication highways for digital businesses.

Code reviews, documentation, meetings with clients, and ensuring effective delegation of tasks between developers and managers were key aspects of my role.

 Some days were intense, but solving technical puzzles was incredibly satisfying ( my dopamine ) .

The best moment? Seeing a client realize the digital solution we created actually transformed their business. .

Want to chat more about tech, hosting, or just catch up over coffee? dm

Head of Software Engineering

Freelancer.com

Do you remember the first dollar you ever made online?

For me, it was $70. I built a web-scraping bot for someone, and let me tell you—the dopamine hit was real. It felt like I had cracked the code.

That $ turned into something bigger.

While I was still a student, I built a freelance business offering machine learning, Python scripting, and web scraping services. Over three years, I didn’t just survive—I thrived, earning more than $30,000 working with clients from all over the world.

I wasn’t just completing gigs; I was mastering skills:
➛ Developing and deploying bots (Discord, Telegram).
➛ Building scalable backend solutions with MongoDB, Docker, and APIs.
➛ Managing cloud deployments on AWS and GCP.
➛ Crafting Python scripts for automation and data scraping.

What started as small projects became a full-fledged business.

I worked with startups, students, and international companies, delivering solutions that made an impact. And through it all, I became financially independent—all while balancing my studies.

That first $70 was just the beginning. But the confidence, the independence, and the skills I gained? That’s the real win.

Blockchain & ML Freelancer

CIT INPT

Member In The Algorithmic and Technical cells at Club CIT INPT

Full Stack Engineer @KohortPay | Building the #1 social payment platform | AI & Automation Specialist

From full-stack developer to AI Agent Orchestrator ( how? 👇)

OpenAI released GPT-3.5 and  that changed everything.

No more endless coding days.
No more slow projects.
No more unhappy clients.

Instead:

— I built the first GPT on WhatsApp. ( Dad is so lazy to go to the web )
— I launched several AI-powered SaaS. ( bevinzey, sanichat, seekgpt...)
— I automated repetitive tasks.

But that was just the start.

While I was working, something unexpected happened:

People kept reaching out. Again and again.

“Can you help me with AI?”
“I need to automate my business.”
“Can you build an AI agent for me?”

I realized I had something valuable.

With the encouragement of David Arnaux, I decided to quit my job and build the best agency for AI agents.

So I did what any smart entrepreneur would do:

I started an agency with my friend Adel.

How do we work?
	1.	In one hour, we analyze your business.
	2.	We figure out your tech setup.
	3.	Our AI agents check your workflows.
	4.	We find weak spots and new opportunities.
	5.	We send you a detailed report.

What happens next?

You decide what to improve.

An AI agent to manage leads?
An automated system for daily tasks?

We handle it.

If we already built it, it’s quick and affordable.
If it’s custom, we give you a fair price.

But be aware:

We’re not like other agencies.

— No overpriced services.
— No empty promises.
— No generic solutions.

We charge less than most.
Why?

Because our AI agents are built for large businesses.

They work faster.
They work better.
They work 24/7.

Are you ready to upgrade your business?

Ready to use AI to move ahead?

Don’t get left behind while others grow.

Reach out today.

Let’s build something great together. 

See you in my calendar: https://cal.link/RqRP1PF

3 hours. 5 full UIs. all built with gemini 3 inside aistudio.
i recorded everything, and i even added a 30-minute breakdown of the exact process that lets us do this every time.

we used to spend days building one website, fixing spacing, correcting layout issues, moving blocks around, and trying to make everything look intentional. now the model creates designs that look like a full design team worked on them for a week straight.

the first one was ayautomate, our studio site. clean sections, strong structure, smooth spacing, and a layout that feels thought out from top to bottom. no template feel. no copy-paste style. the ai got the visual language right on the first attempt.

then aytalents, which is easily the strongest design we’ve made so far. long flowing sections, perfect spacing, balanced typography, and a look that feels original. this is the kind of page that normally needs multiple rounds of refinement. here it came out complete in minutes.

after that came aybrowser. one prompt produced a full landing page with a hero, feature sections, pricing layout, and a simple testimonial area. the look is clean, modern, and well aligned. the structure makes sense from the first block to the last.

then ayevents, which came out with a futuristic look. timeline segments, speaker cards, access levels, structured blocks, and a long layout that still feels readable. the theme is clear across every part of the page. the ai understood the style and maintained it consistently.

the final one was the visual suite page. strong visuals, stats, product sections, and a layout built for creative tools. the spacing is tight, the shape is clean, and the overall look outperforms what we used to build manually.

i didn’t edit icons
i didn’t fix spacing
i didn’t reorganize the layout
the ai handled everything from the first prompt
and the results look like custom work done by hand.

▶︎ i also recorded a full 30-minute walkthrough showing the full method that lets us build ui like this fast.

▶︎ you can watch every design, the full flow, and how we prompt to get this level of quality.

🔥 Want the full system?
❤️ Like this post and comment ui and i’ll send it to you.


---
we will launch our new offer next week.
if you want to reserve a spot and get the best ai/n8n engineers in your team, reach out before we launch it publicly.
the price already jumped two times in three months.

Walid Boulanouar

building more agents than you can count | aiCTO ay automate & humanoidz | building with n8n, a2a, cursor & ☕ | advisor | first ai agents talent recruiter

connect n8n to claude code. ( insanely good 🔥 )
i never thought about it seriously until i tried it once, and it changed how i use both tools forever.

i always liked n8n, but it had limits.
i always loved claude code, but it lived alone in the terminal.
the moment i connected them, everything clicked
it felt like giving n8n a real engineer inside my server.


i made a 20 min video breaking down the exact setup. the ssh connection, the session id trick, the sub workflows, the agent tools, the full flow from idea to execution. simple and practical.  if you want it, comment “ssh”.

sometimes,from slack or telegram i can talk to claude code like an assistant who already knows my machine. check this, fix this, inspect that, run the script, debug the project, deploy the update. it does the work and n8n orchestrates the pipeline.

session id keeps context alive. ( i can even scale it to multiple claude code instances based on the users)
so the agent feels persistent, not stateless.
it’s fast, predictable, and way more capable than any rag setup i tried for system tasks.

and the same setup works with gemini cli or codex.
you can even plug github into the loop and trigger workflows that update projects automatically.
your server becomes the workspace.
n8n becomes the conductor.
claude code becomes the engine.

i wish i had this blueprint months ago.
it would have saved me weeks. 
to get it

1) ❤️ like

2)💬 comment “ssh”

3)🔄 repost (if you want)

→ more deep dives coming next.

( thanks nano banana for the visual )



edit: you can get the resource from this link -> https://lnkd.in/ePXuqjhh

i just vibe-coded a football game with gemini 3
nano banana pro handled the visuals and it looks more real than real life
what a time to be alive 🤯

if you want it comment so i can send it in 2030 D

end your week with this....  a lot of creativity ✨

made by : Unreal engine + after effects for visuals. Madmapper for projecting

credits: https://lnkd.in/ehr8ExhY

am I missing something? 🫠 

cc: Google, Qoder, Cursor, Windsurf, Kiro, Google Antigravity, Browserbase, TRAE, Visual Studio Code, Perplexity, Strawberry, OpenAI

drop any file you have and see it brought to life.
that's what i built in gemini simply by using 12 prompts.

long story short
i spent the last few days inside aistudio building a bunch of small tools.
one of them ended up becoming the most fun thing i’ve made in a while.

i call it anything.
you drop a pdf
an image
a video
or even a long document
and gemini 3 brings it to life in seconds.

it reads it
understands it
breaks it down
and turns it into something you can work with.
games, small tools, visuals, an app yes an app that can build apps.

i submitted a picture i made with nano banana of me in the white house with elon and trump and it made an app like i asked.
i submitted a screenshot of a 2d football game and it built it.

the craziest one is this
i dropped an invoice picture and it built me a custom invoice generator that worked pretty well 🫠

you can build on top of it by accessing the link in the comments.

also you can find a 16 minute video where i showed what i built like a clone of gamma, arcads, a thumbnails generator, and more.

weekend mode: building things nobody asked for, as usual.

i am thinking of making a series where each day i build something useful in public I would like to know if you are interested in that kind of challenge

someone rebuilt their entire simulator UI using nano banana + gemini 3
and the glow up is insane.

this is exactly the kind of stuff i love to see
because it’s the same jump we experienced when we started using nano banana inside aistudio for our own builds.
you take an old interface
screenshot it
pick a style
and the ai gives you something you would never design by hand.

the flow is simple:
→ screenshot your current UI 
→ nano banana gives you clean variations in seconds
→ choose the vibe you want
→ send the layout to gemini 3 to turn it into real code
→ iterate again without losing time or budget

results look like a full redesign made by a team
but it took minute
and you can change themes anytime.

i rebuilt entire pages this way last week; ayautomate, aytalents, aybrowser, ayevents,ayvisual suite, plus viralbrain that is coming.
each one came out cleaner and more polished than our old manual designs.


this combo is becoming the secret weapon for builders who want to ship fast.
prompt → pick → polish → deploy.
i already recorded 52 mins on how i did it with bunch of prompts, steps and miro...
I share it with more than 400 people commented on my post
if you still not get it yet drop a comment and i will send it to you.

will i stop using openai?
honestly… if gemini adds a way to build custom gpts with full mcp support, it becomes a real question.

anyone here tested something close to that?
curious to hear real experiences.

we just launched the new ayautomate branding, released ayn8n, and shipped ayclaude.
this week we’re finally putting out aytrust. it’s been ready for two weeks, we just delayed it.

we’re also launching viralbrain, our internal tool that helped me grow to 16k followers in six months.

yes, the same system we use every day to write, test, and scale posts that work. (cc: David )

and we have a few more releases coming at the start of the year.
if you want something specific, tell me.

put your actual pain in the comments
and we’ll build the tool that solves it for you.

paris metro. 1 hour sitting. why not build something using Replit quickly and dirty.

i’ve had this idea in my notes for a long time.
a simple setup where you don’t chat with the ai
you let two different models talk to each other.
that’s the whole point.
i finally built the first version today.
i call it aychat.

the idea is this.
you pick model A
you pick model B
for example gpt-5.2 on one side
claude sonnet 4.5 or  gemini 3 on the other
and you let them talk.
one pretends to be the “user”
the other replies like the assistant
and the full conversation grows without you typing anything after the first line.

so you send one message to start the scenario + shared context and system prompts.
and they take over.
model A creates a challenge
model B tries to solve it
then model A checks the answer
and sends feedback
and they keep going.

today i tested it with a puzzle setup.
model A invented a hard puzzle
one he didn’t know the answer to
model B tried to solve it step by step
and model A came back to review the solution
explain the logic
and push the next question.

you get a full debate
two different ai personalities
two different ways of thinking
and you just watch them work through it.

i built the prototype inside Replit while the metro was stopping every few stations.
quick build
rough edges
but the idea works.

good idea??
or just kill it...

let me know in the comments

last weekend i was running a workshop with my team on how to build a second brain using claude code, and how to connect an agent to it using n8n.


we ended up building a system that lets us control a remote server holding our entire second brain - company processes, sops, templates, internal resources… everything. 

honestly, this is a killer feature that i would recommend to any team (yes, really).
the setup is actually simple. 


you connect n8n with claude code using the tutorials i shared in one of my posts. if you want the resources, just comment 👀. then on top of that, you can use slack or telegram to build a bot to interact with it, and even connect it directly to your agents (this part is fun).


this opened a lot of possibilities for me and for the team. 
we are already saving a lot of time, and the way we operate feels very different (in a good way).
we will add this workflow to ayn8n tomorrow, and i will drop the link in the comments .


building a second brain that your entire company can access is probably one of the best things i did this year. maybe that sounds exaggerated 😅, but i can clearly see how big of an unlock it is for how we work.


if you want me to build a full playbook around this, let me know in the comments how you are doing it on your side. i am also happy to open source my second brain github repo (once i clean it a bit 🙃).

it still needs some polishing, but there is already a lot of value here.
and with the new n8n drop, i am excited to see what we will build in the next workshop with the team

hey friends
please do not overthink my gift
simple things make me happy 😌

my birthday is in march
but i am flexible
we can move it to today if that helps with gifts 

new year gifts for people who love AI and happiness 

1. Claude AI gift card - claude.ai/gift from 18 euros
  this one says I care and I actually read docs

2. Perplexity Pro gift card - https://lnkd.in/eqDUY-4N from 20.95 euros
  this one says I hate Google but in a respectful way

3. Gemini AI and ChatGPT
  no gift cards
  missed opportunity
  please fix this next year 🎁 ( i need gemini max max plan ) 

thank you
I will remember this forever
or until the next model update

this is how i manage my second brain from telegram/slack 🔥

i just uploaded our workflow for managing a second brain from a telegram bot

you can adapt it to any messaging platform( slack, teams.... ) 

1) open ayn8n
2) click newest first
3)open the workflow and click copy workflow

then follow the detailed steps using the 30 minute video i recorded
you can find the link in the workflow description


enjoy the most productive setup you will ever use

gpt-5.2-high-x-low-reasoning-mini-intense-no-codex-medium-short

GPT 5.2 (none)
GPT 5.2 Low
GPT 5.2 Medium
GPT 5.2 High
GPT 5.2 Extra High
GPT 5.2 Pro Medium
GPT 5.2 Pro High
GPT 5.2 Pro Extra High

GPT 5.1 Codex (none)
GPT 5.1 Codex Low
GPT 5.1 Codex Medium
GPT 5.1 Codex High
GPT 5.1 Codex Mini (none)
GPT 5.1 Codex Mini Low
GPT 5.1 Codex Mini Medium
GPT 5.1 Codex Mini High
GPT 5.1 Codex Max (none)
GPT 5.1 Codex Max Low
GPT 5.1 Codex Max Medium
GPT 5.1 Codex Max High

the next update of claude code or cursor won't save your app, these 30 tips will
after debugging a few ai-heavy apps, this is what actually worked for us:

you're coding in 2025
you're using claude, gpt-5.2, cursor, gemini, openrouter
not a typewriter
so build like it

here's 30 tips that saved us weeks of guessing and thousands of wasted tokens:

1. lock model behavior in .rules, or .claude files across tools
2. keep max 5-10 open files active - more = token chaos
3. use exact paths or symbols in prompts to scope ai attention
4. never write code before a step-by-step task plan
5. let claude plan first, then code - don’t skip the blueprint
6. paste api docs in prompt - hallucination drops instantly
7. test-first dev works better than generate-then-fix
8. claude isn’t confused - your context is
9. describe the problem before the solution - structure matters
10. rerank retrieved chunks - don’t trust raw similarity scores
11. use xml or json tags for structured claude responses
12. reset sessions before 90k tokens - context decay is real ( put a memory.md file or handoff.md file to put inside it the full summary)
13. add human-readable comments - they teach future ai runs
14. use # notes in claude to correct and train persistence
15. end sessions with ai-generated testing plans
16. use visual and code agents together for fast ui feedback
17. commit code often - git is your memory
18. let claude write variations - pick the cleanest
19. restrict model scope - "only use this context" works
20. plan first, prompt second - always
21. don’t inject 20 pdfs - inject 3 that matter
22. switch models by task - gemini for scan/taste, claude for logic
23. write your constraints before asking for code
24. scan your codebase weekly for broken context links
25. slow claude = too much junk in memory - clean it
26. store memory in .claude not in 40 prompts
27. debugging hallucinations? reset and reframe the problem
28. clarify your input formats - don’t assume the model will guess
29. sync gemini structure maps into claude rules
30. claude isn’t your dev - it’s your noisy intern. train it like one

steal these. screenshot them. drop them in your rules file( or sop for your team ) 
don’t wait for better updates
write better instructions

new chatgpt image model 𝘃𝘀 nano banana pro model

google really played the game well ahah

they got the codebase behind windsurf
they got the whole windsurf team
what a deal google

they even got the founder of character for like 2B i think ( 1 year ago, when everyone was mocking on google models)
and now they are crashing all benchmarks

buy what works
don’t try to reinvent the wheel
i thought it was just us doing this ahah

gemini 3 pro + claude code inside google’s new antigravity ide?

what i can say is it’s meh.

it looks like it’s doing magic, but the ux has many issues. it doesn’t make me feel like i’m coding in a good environment.

taplio

Walid Boulanouar's Linkedin Analytics

Walid Boulanouar

Performance ScoreWalid Boulanouar's LinkedIn Performance

Walid Boulanouar's Best Posts (last 30 days)

Ready to grow your LinkedIn presence?

taplio

.css-6ytdur{background-image:linear-gradient(284deg, rgba(39, 43, 72, 1) 0%, rgba(33, 47, 183, 1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Walid Boulanouar's Linkedin Analytics

Walid Boulanouar

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Walid Boulanouar's LinkedIn Performance

Walid Boulanouar's Best Posts .css-1emoo84{font-size:var(--chakra-fontSizes-md);font-weight:400;color:var(--chakra-colors-gray-600);}(last 30 days)

Ready to grow your LinkedIn presence?

Walid Boulanouar's Linkedin Analytics

Performance ScoreWalid Boulanouar's LinkedIn Performance

Walid Boulanouar's Best Posts (last 30 days)