Chibuike Mba

A software architect/engineer who loves analyzing, designing and implementing software solutions with varying complexities.

[Chibuike Mba @chibex]

A good system shouldn't pretend to understand.

It should:

→ Clarify → Ask for more information → Or hand over to a human
AI knowing its limits is part of intelligence.
We're thinking a lot about this while building Flowk.

#AIEngineering #SoftwareEngineering #SystemDesign

[Chibuike Mba @chibex]

The impressive AI demo is when everything works perfectly.

The real engineering challenge is:

What happens when it doesn't?

Noisy voice notes.

Blurry screenshots.

Incomplete information.

Unexpected customer behaviour.

[Chibuike Mba @chibex]

The challenge isn't just generating a payment link.

It's understanding when the customer is ready for the next step.

Context should guide the conversation.

[Chibuike Mba @chibex]

A customer asks questions.

You answer them.

They decide to buy.

Then:

"Please go to our website and complete your purchase."

Sometimes that's necessary.

[Chibuike Mba @chibex]

But it raises an interesting question:

Why should the customer journey feel disconnected from the conversation that created the buying intent?

That's something we're exploring with Flowk.

#ConversationalCommerce #Flowk

[Chibuike Mba @chibex]

The difficult part isn't just processing voice or images.

It's knowing what to do when the AI isn't confident:

→ Ask for clarification
→ Request more information
→ Bring in a human

[Chibuike Mba @chibex]

Your customers don't always want to type.

They send:

🎙️ Voice notes
📸 Screenshots
🖼️ Product images
🧾 Payment receipts
But many customer systems still basically say:
"Sorry, I only understand text."

[Chibuike Mba @chibex]

I think software should adapt to how customers communicate — not the other way around.

That's one of the problems we're thinking about with Flowk.

[Chibuike Mba @chibex]

Our philosophy is simple:

AI handles the predictable conversations. Humans handle the conversations that need humans.

We're still building and learning.

[Chibuike Mba @chibex]

5 signs your customer inbox may be costing you revenue:

Customers wait too long for answers.
Your team answers the same questions repeatedly.
Customers keep repeating themselves.
AI has no clear path to human handover.
Support becomes harder to manage as you grow.

[Chibuike Mba @chibex]

These are some of the problems we're building Flowk around.

Which one do you see most often?

[Chibuike Mba @chibex]

The difficult part isn't getting AI to answer.

It's deciding what should happen next.

[Chibuike Mba @chibex]

A good conversational sales flow shouldn't feel like a chatbot.

It should:

Understand the customer's context
Understand their intent
Give them useful information
Know when to bring in a human

That's the thinking behind Flowk.

[Chibuike Mba @chibex]

AI handles the predictable.
Humans handle what requires judgment.

#SalesAutomation #ConversationalAI #Flowk

[Chibuike Mba @chibex]

AI can then handle the repetitive part of the conversation while a human steps in when needed.

We're building around that AI + Human model.

[Chibuike Mba @chibex]

You paid to get the customer into WhatsApp.

Then your business asks:

“Hi! How can we help you?”

The customer already told you what they're interested in by clicking the ad.

The problem isn't just getting the lead.

It's carrying the context into the conversation.

[Chibuike Mba @chibex]

That's one of the problems we're exploring with Flowk.

[Chibuike Mba @chibex]

AI can now do part of the work.

So why are we still primarily pricing software by the number of humans using it?

That's a question we're exploring while building Flowk.

**If AI does the work, shouldn't pricing reflect the work?**

[Chibuike Mba @chibex]

Your customer doesn't care that your support team is busy.

They only know they sent your business a WhatsApp message...

and nobody replied.

Response time isn't just a support metric.

When you're selling through conversations, it's part of the customer experience.

[Chibuike Mba @chibex]

Your customer doesn't care that your support team is busy.

They only know they sent your business a WhatsApp message...

and nobody replied.

Response time isn't just a support metric.

When you're selling through conversations, it's part of the customer experience.

[Chibuike Mba @chibex]

Most companies are asking the wrong question about AI customer support.

They're asking:

“How many support agents can AI replace?”

I think the better question is:

“How many more customers can my team serve if AI handles the repetitive conversations?”

[Chibuike Mba @chibex]

@Nada_kln There are several good options depending on what you want to automate.

We're actually building one ourselves — AI handles the repetitive stuff, humans step in when it matters.

More on that soon. 👀

[Chibuike Mba @chibex]

Most companies are asking the wrong question about AI customer support.

They're asking:

“How many support agents can AI replace?”

I think the better question is:

“How many more customers can my team serve if AI handles the repetitive conversations?”