
Advocating for @googlecloud | Mentorship PM ~@techrityorg Facilitator&Tech Trainer @SpitalLabs | organising @gdgcloudportharcourt -I ❤️ travel & tech #Mademind
Innovators' Planet
Joined November 2014
- Tweets5.5K
- Following1.2K
- Followers553
- Likes1.8K
You don't need code to build software.
Vibe coding explained in 5 mins: what works, tools and limits.
Watch the full video here 👇
youtu.be/eLnSuREaIcA
Why did Google build Gemini 4 Argon and lock the doors? The truth behind its private release:
youtu.be/ajFilZLakkY
🤖 Made with AI
Caleb Ini. D. Duff | TLGM | retweeted
A warning to all designers
I'm seeing a dangerous trend on LinkedIn and X, and I need to call it out.
In trying to prove you can "design and build with AI," designers are oversimplifying the process. And it's backfiring badly.
What I'm seeing:
"I built an entire app in seconds with AI."
"You don't need to give any input, AI handles everything."
"Designed and shipped in hours with no effort."
Imagine you're a potential client reading that. Why would you hire a designer if it sounds like they did nothing? Like AI did all the work?
Here's the reality:
Anyone who's actually built something with AI knows the truth. The work is immense.
You need to:
- Write detailed PRDs and briefs
- Know which AI tools solve which problems
- Run commands in the terminal
- Manage API tokens and costs
- Make countless design iterations
- Test, refine, validate
- Make critical decisions at every step
This isn't automation. This is strategy disguised as simplicity.
The damage you're doing:
By making it sound effortless, you're training clients to think designers are obsolete. You're undercutting your own value.
What you should actually communicate:
"I used AI strategically to accelerate execution while maintaining my standards. Here's the thinking behind the decisions. Here's what I refined. Here's the work that went into this."
That's the message that gets clients excited. That's what justifies the rate.
Remember:
Your critical thinking is still essential. Your judgment still matters. Your ability to see what AI missed, that's irreplaceable.
Don't sell yourself short trying to sound impressive. Show the actual work. Show the thinking.
That's what makes you valuable.
2/2
Define an agent in ~10 lines of Python, deploy with one command, then pick your inference mode:
Live user waiting? → online
100k tickets overnight? → batch
Full walkthrough ↓
dev.to/gdg/managed-inference…
#GoogleCloud #CloudRun #AI
Shipping AI apps on Google Cloud without touching a GPU
The pattern: Gemini Enterprise Agent Platform handles orchestration, memory & models. Cloud Run hosts your code, tools & MCP servers.
Each tier scales — and fails — independently.
1/2
#GoogleCloud #CloudRun #AI
How do you let AI agents execute dynamic, untrusted code without putting your infrastructure at risk?
In my latest article, I break down Google Cloud Run Sandboxes.
Read the full breakdown on DEV Community 👇
dev.to/duffigoogle/cloud-run…
Google has introduced two additional capabilities for the HA Serverless Cloud Run service for a smooth implementation of Cross-Region Failover and Failback.
dev.to/gdg/zero-to-multi-reg…
This brings memories; the waves, the flinging from the boat seats, the engine stopped, the panic.
I can still remember praying silently on one of those trips ... "Peace be still over this water body", "There is peace within my walls/territories".
Have a good read about VPC design in production level.
medium.com/@duffdev001/what-…





