Last week, I almost became a customer. Almost.
A SaaS company in Indonesia (can't really say it because i feel bad for them lol), had a product that did exactly what I needed. WhatsApp automation with interactive forms, multi-step menus, and lead collection. Their pricing page said Rp400,000 per user per month. Reasonable. I registered.
Then reality hit.
What I Wanted
Quick context: I run Kugie, a small digital solutions company. We needed a way to automate lead collection through WhatsApp -because in Indonesia, WhatsApp isn't just a messaging app. It's the platform. If your business can't handle WhatsApp conversations, you're leaving money on the table.
My inspiration was tiket.com's customer support -Halo Tiket. You message them and get this slick experience: menus, options, forms -all inside WhatsApp. No links to external pages. No clunky workarounds. Just a smooth, native flow.
I wanted that for Kugie.
The Quote That Changed Everything
After registering, a sales rep contacted me. We had a solid conversation. I even shared a screen recording showing exactly what I wanted to build. She suggested a demo for the following Monday.
Cool, I'm in.
But then she sent the quotation early. And buried in it was one line that changed my entire plan:
"Minimum 5 users."
Five. Users.
We're a tiny, agile team. I don't need five seats. But the plan requires it.
Rp400,000 × 5 = Rp2,000,000/month.
I closed WhatsApp and tried to forget about it. I had dinner plans with friends.
I could not forget about it.
Over dinner, I kept doing the math in my head. I even told my colleague -"Can you believe they require minimum 5 users? That's Rp2 million a month for WhatsApp automation." He was just as baffled.
The whole evening, it was stuck in my brain. Five users. Two million rupiah. For something that felt buildable.
I got home, sat down, and opened Claude.
3 Hours Later
"Hey Claude, can I build WhatsApp interactive forms using Meta's Business API?"
What I love about Claude's project memory is that it already knew Kugie's context. It knew we run a product called Swivel that automates WhatsApp for our clients. So its first response wasn't just "yes, it's possible" -it was "yes, and you can probably reuse the WhatsApp Business number you already have."
That was the unlock. I didn't need new infrastructure. I just needed to extend what we already had.
Here's what the next 3-4 hours looked like:
Discovered WhatsApp Flows. This was completely new to me. Meta has a feature that lets you build multi-step, interactive forms that run inside WhatsApp. Native UI, no webviews. The approval process was surprisingly painless.
Created message templates. Connected the flows to WhatsApp message templates so they trigger automatically when someone reaches out.
Built a Hono endpoint. I needed a lightweight webhook handler to process inbound messages from Meta's API. Considered Express (too heavy), Next.js (overkill), and Go (too much ceremony for a simple handler). Landed on Hono -minimal, fast, and familiar.
Connected everything to Sleekflow. Our team uses Sleekflow for messaging, so the endpoint forwards processed messages there. Everyone can see and respond to leads.
Deploy. Test. Done.
What I Actually Saved
It's not just the money -though let's be clear, Rp24 million per year isn't nothing for a small company.
It's the dependency. With a SaaS tool, I'm locked into their feature set, their pricing changes, their roadmap. With my own solution, I control everything. Need to add a new form field? I change it. Need to route leads differently? I update the endpoint. No support tickets, no waiting for features.
And the whole thing is maybe 200 lines of code.
On "AI is Killing SaaS"
Every week I see another post on X about AI replacing SaaS products. I used to scroll past them.
Now I am one of those posts.
But here's the nuance that most of those takes miss: AI didn't replace the SaaS by generating code. The code was the easy part. What AI actually did was compress my research time.
Before Claude, here's what this project would have looked like:
- Google "WhatsApp automation API" -wade through SEO-optimized garbage
- Find Meta's docs -get lost in their sprawling documentation
- Discover WhatsApp Flows exists -maybe, after hours of reading
- Figure out I could reuse our existing number -probably not, without someone telling me
- Pick a framework, scaffold it, debug it -the usual
That's easily a week of scattered work. Maybe two if I kept getting pulled into other tasks.
With Claude, the discovery-to-deployment pipeline collapsed into a single afternoon.
That's the real disruption. Not "AI writes code." It's "AI eliminates the research phase that used to make small projects feel too expensive to start."
For Small Teams in Indonesia
If you're running a lean startup or a small agency here, this is my honest advice:
Before you sign that SaaS contract, spend an hour asking Claude (or whatever AI tool you prefer) whether you can build it yourself. You might be surprised.
Not everything should be DIY -I happily pay for plenty of tools. But for straightforward automation workflows? The build-vs-buy calculus has shifted dramatically.
The tools exist. The APIs are free (or cheap). The AI will walk you through the parts you don't know.
You just have to ask.