Introducing ArtiSANs — Nigeria's first verified artisan marketplace.
We're solving a problem every Nigerian knows:
You need a plumber.
You call 3 numbers.
One doesn't pick up.
One shows up late.
One overcharges.
There's a better way. 👇
@MTNNG@MTN180 Let this be a very clear warning. Stay away from my line. If you are looking for clients to steal data from, just remove my number. Take this as the last warning as I might be the one to lead the protest to your HQ.
Be warned!!
Thanks.
- ArtiSANs lets you POST a job and receive COMPETITIVE BIDS from verified artisans in your area.
- Every artisan is verified with NIN/BVN. No more "my guy" who disappears after collecting payment.
- Compare prices, read reviews, choose who to hire.
Introducing ArtiSANs — Nigeria's first verified artisan marketplace.
We're solving a problem every Nigerian knows:
You need a plumber.
You call 3 numbers.
One doesn't pick up.
One shows up late.
One overcharges.
There's a better way. 👇
Arsenal are officially Premier League Champions. Ive waited 22 years to say that. Thank you Mikel Arteta. Unforgettable’s. What a journey it has been with each and every one of you. Champions of England. Easily one of the best days of my life. We did it ❤️
You are cursed o. Proper cursed.
Multiple kidnappings and abductions from schools, you have no business with those children.
You're staggering through party politics.
You're a demon o. A proper demon.
Can't wait for the day we don't have to bear your ugliness anymore.
How I have bee in building with my Hermes agent. I have a workflow repo where I define the workflow for performing any task. My workflow always includes a security check to ensure that the agent is building a secure system for me always.
Using one AI agent works… until it doesn’t.
The moment you add multiple steps — things break.
So instead of “using AI,” build a simple system: Request → Plan → Change → Preview → Approve → Monitor
It turns AI from unreliable → controlled.
Here’s exactly how it works
I agree with this.
I’ve heard a lot about taking a course in prompt engineering but overtime I’ve come to understand that prompting the AI when you know exactly what you want , how it should be done, and what to expect makes building with AI easier.
People have been asking me this.
Honestly, the way I vibe-code is not special at all. Going by the true definition, it's not even vibe-coding.
The reason I build good software with AI is because I was already good at it prior to AI.
I don't even use skills, agents.md, MCPs, etc.
I prompt one small feature at a time.
My typical vibe-coding session looks like this:
"Create a table to track usage analytics, the data should be scoped to projects for better granular insights."
> Agent creates table.
"Generate and run the migration files, then create a trackUsage helper function we can resuse".
> Agent: Done.
> "Scan the API directory and track usage for every event that is emitted".
Sometimes, I send all of this or more as one prompt but you get the point. One thing at a time, the same way I would do it without AI.
I'm not doing anything out of the ordinary to get good results out of AI. Existing experience and expertise is what helps me.
A Nigerian fintech just lost ₦20 million to a fake webhook.
Attacker didn’t hack anything.
They just POSTed this to the endpoint:
{
"event": "transfer.success",
"amount": 500000,
"status": "success"
}
Backend credited the user. Zero money moved.
This is happening more in crypto payments too.
What every backend dev must do in 2026:
1. Verify webhook signature + IP + timestamp (not just event name).
2. Never credit on webhook alone always confirm on-chain + NIBSS.
3. Add rate limiting and replay attack protection.
4. Reconcile every stablecoin inflow against blockchain truth.
Crypto rails move fast. One lazy endpoint and you’re done.
Save this like your production depends on it.
Drop your worst webhook horror story 👇