How We Built Apaya: The $15,000 Monthly Mistake That Led to Our Next Company

“Necessity is the mother of invention. Desperation is the father who actually gets off his ass and builds it.”
– Me, after our third all-nighter doing social media

I was sitting in our home office at 2 AM, manually scheduling Instagram posts about toddler education while my co-founder Vivek was in Canva trying to make cartoon vegetables look “engaging.”

This was peak startup glamour. Two guys who’d previously sold a company for decent money, reduced to copy-pasting hashtags about preschool learning at ungodly hours.

“We’re spending more time on marketing than product development,” Vivek said, not looking up from his fifteenth attempt at a Facebook ad graphic.

He was right. And that’s when I knew we were completely screwed.

The $15,000 monthly disaster that was eating Kokotree alive.

Let me paint you the full picture of our marketing “strategy” at Kokotree (our educational app for preschoolers):

Option 1: Do It Ourselves

• 30 hours per week between us
• Posting sporadically when we remembered
• Content that looked like dads trying to be cool
• Zero consistency, zero strategy
• Opportunity cost: $15,000/month in lost product development time

Option 2: Hire an Agency

• $8,000/month for “full service”
• Generic posts about “the importance of early learning”
• They kept calling our app “KokoTree” (one word, wrong capitalization)
• Weekly calls where they’d present our own ideas back to us
• Results: 12 likes per post, mostly from our moms

Option 3: Build a Marketing Team

• Social media manager: $65,000/year
• Content writer: $55,000/year
• Designer: $60,000/year
• Total: $180,000/year plus benefits
• For a pre-revenue startup. Yeah, right.

We tried all three. Failed at all three. The agency lasted two months before I fired them in what Vivek calls “the most polite rage quit in startup history.”

Why every existing solution was built for companies that aren’t us.

Here’s what pissed me off most: Every marketing tool assumed we had unlimited time, money, or both.

Hootsuite wanted us to create all the content first, then they’d graciously post it for us. Thanks for solving 5% of the problem, guys.

Buffer was the same thing with a prettier interface. These schedulers are just expensive alarm clocks.

The AI writing tools? They’d generate content that read like a robot having an existential crisis. “Dear valued parents, we are pleased to announce the availability of educational content for your offspring.” Kill me.

Nobody understood the actual problem: We needed the entire marketing department, not just one tool.

The midnight epiphany (Or: When desperation meets capability).

It was a Thursday night. I’d just spent four hours creating one week of social media content. Four hours. For seven posts. That would probably get 50 views total.

“This is insane,” I told Vivek. “We built a collaboration platform that handled millions of users at Axero. We’re smart enough to build educational apps. Why the hell are we doing this manually?”

“Because nobody’s built what we actually need,” he said.

“Then let’s build it.”

“We’re already building Kokotree.”

“So? We’ll build both.”

The look he gave me suggested I’d lost my mind. Maybe I had. But sometimes the best ideas come from the intersection of fury and exhaustion.

Building the marketing system we desperately needed.

We didn’t set out to create another company. We just wanted to stop drowning in marketing tasks. But once we started building, we couldn’t stop.

Week 1: The Brand Learning System

We built an AI that could crawl our Kokotree website and understand our brand. Not just scrape keywords—actually comprehend what we were about. Our voice, our mission, why we existed.

This became our AI Brain—the foundation that ensures content sounds authentic, not artificial.

Week 2-4: The Content Generation Engine

We trained it to write like us. Fed it our best content, our worst content, and everything in between. Taught it the difference between “engaging parents” and “sounding like a PBS documentary from 1987.”

The breakthrough: Using LLMs that actually understand context, not just string words together.

Week 5-8: The Visual Design System

Every post needs graphics. We were burning hours in Canva. So we built automated design generation that understood our brand guidelines and created on-brand visuals without human intervention.

Week 9-12: The Publishing Orchestra

Connected it all to social platforms. Not just scheduling—intelligent distribution. The AI learned when parents were actually online, not just generic “best times to post.”

Week 13-16: The Learning Loop

Built analytics that fed back into the system. What worked? What didn’t? The AI learned and adjusted, getting smarter with every post.

What happened when we turned it on.

The first week Apaya ran for Kokotree was… educational.

Look at that graph. See early August? That’s us doing social media manually—sporadic posts whenever we remembered, engagement flatlining around 200-300 per week.

Then September hit. We turned on Apaya.

Week 1: Engagement doubled to 400+
Week 2: Climbed to 600+
Week 3-4: Stabilized around 650+

Here’s the thing though—those early Apaya posts were trash. Pure AI slop. They looked like a robot’s fever dream about marketing. Generic. Soulless. The kind of content that screams “I let ChatGPT write this.”

But even our garbage AI posts were better than no posts! We were thrilled!

And here’s the beautiful part: The system kept learning. Every week it got slightly less terrible. The posts went from:

• “Embrace the power of early childhood education!” (kill me)
• To actual useful content about preschool development
• To posts that finally sounded like humans wrote them

The numbers from back then:

• Posted daily instead of “whenever we remembered”
• Reach grew from 7K to 16K accounts per week
• Engagement tripled despite the mediocre content
• Zero hours spent on social media

What would happen if we turned on today’s version of Apaya back then?

I honestly think we’d have 10x’d these numbers. The posts we generate now are 100x better—actual valuable content, not AI word salad. We’ve spent months teaching it to not sound like a LinkedIn motivational poster.

Vivek called me after two weeks of the original version: “Our social media looks… consistent.”

“Not good?”

“No, definitely not good. But it exists. And it’s there every day. That’s more than we had before.”

He was right. Even mediocre consistency beat our previous pattern of brilliant posts followed by three weeks of tumbleweeds.

The slight dip at the end of October? That’s when we started tweaking things, proving the AI was doing better without our “help.”

Today’s Apaya would embarrass this early version. But this early version still beat doing nothing—which is what most businesses are effectively doing when they post randomly every three weeks.

Why we’re giving everyone the tool we desperately needed.

Look, I get the skepticism about “another AI tool.” The market is flooded with ChatGPT wrappers and “revolutionary” platforms that revolutionize nothing.

But Apaya is different because we built it from pain. Real, acute, “I’m going to throw my laptop out the window” pain.

We didn’t build features because they seemed cool. We built solutions to problems that were actively destroying our sanity.

Every feature exists because we needed it:

• Brand learning because we were tired of sounding generic
• Automated design because Canva was eating our souls
• Smart scheduling because posting manually is medieval
• Performance analytics because we needed to know what worked

This isn’t theoretical. We use Apaya every day for Kokotree (and a handful of other “fake” businesses I started just to test it). It’s not perfect, but it’s approximately 1000% better than doing it yourself or hiring an agency that doesn’t get you.

The plot twist: Kokotree is thriving (And so is Apaya).

Today, Kokotree is growing. People are signing up. Not because we became marketing geniuses, but because we automated some of the marketing so we could focus on making the product better for our customers.

And Apaya? It’s helping businesses escape the same trap we were in.

The beautiful irony: We built a company to help us build a company, and now both are succeeding.

Your turn to stop the madness.

If you’re reading this at 2 AM while scheduling tomorrow’s posts, I have a message for you: Stop.

If you’re paying an agency that doesn’t understand your business: Fire them.

If you’re considering hiring a full marketing team you can’t afford: Don’t.

The solution we built for ourselves can be your solution too. Because every hour you spend on manual marketing is an hour not spent on your actual business.

Start your free trial and see what happens when marketing runs itself. No credit card required. No committees. No consultants.

Just the tool we wished existed from day one. Now it does.

P.S. – If you’re wondering if two founders building two companies simultaneously is insane, you might like my book, because it’s full of questionably sane business decisions that somehow worked out.