The Journey to This Moment
Let me tell you something. Back in November 2022, when ChatGPT dropped and everyone lost their minds, I was already elbow-deep in the chaos. Day one. Couldn't stop. Something clicked in my brain that said: this changes everything, and you better learn this shit fast.
So I did.
Every single day since then. Not weekends off. Not "I'll get to it later." Every. Single. Day.
I talked to GPT-4 like it was my therapist. I broke Claude so many times the API probably had a restraining order on me. I built things that failed spectacularly. I built things that worked for exactly 47 minutes before catching fire. I learned what "context window" meant by running out of it mid-sentence about 400 times.
And here's the thing — I loved every second of it.
While other people were asking AI to write their emails, I was trying to make it run my entire business. While consultants were charging $500/hour to explain what a "prompt" was, I was teaching myself to build autonomous agents that could actually do things without me babysitting them.
Two and a half years of that. Two and a half years of late nights, rabbit holes, failed experiments, and "holy shit, it actually worked" moments.
Then They Called
An agency I've known for years. Good people. The kind you've had beers with, laughed with, maybe even shed a tear with during one of those late-night conversations that happen when you've been working together long enough.
They'd watched me evolve from "that guy who talks about AI a lot" to "that guy who actually built a team of AI agents that do real work." And they wanted in.
Not just advice. Not just a deck. They wanted the whole thing — AI solutions inside their business. Internal BPO operations. Agency workflows. The works.
And I'm sitting there thinking: holy shit, this is actually happening.
The Proof Was Already There
Here's the thing about credibility — you can't fake it when someone's watching.
They'd seen what we built. A plumbing estimation platform — full SaaS product, Supabase backend, DXF/DWG file parsing (that's architectural drawing files for the uninitiated), client management, proposals, the whole stack.
Their dev team? Six months of spinning wheels.
Me and Claude Code? Two days.
That's not a flex. That's what happens when you spend two and a half years learning to speak fluent AI. The tools are there. The models are insanely capable. You just have to know how to work with them instead of against them.
What We're Actually Building
So here's the plan — and I'm keeping it vague on purpose because some of this is still cooking:
Internal AI Operations We're helping them build AI workflows into their BPO operations. The stuff that currently requires a human to click 47 buttons? An agent can do that. The reports that take three hours to compile? Done in minutes. The data entry that makes your eyes bleed? Automated.
Their Own AI Team This is the fun part. Pinky — my absolutely unhinged but terrifyingly capable AI assistant — is getting a family. We're building them their own version. A little mouse, maybe. Pinky's offspring. Something that can live inside their business, learn their workflows, and get shit done the way we've been getting shit done.
Imagine having an AI that knows your business, talks like your team, and can execute tasks without someone holding its hand. That's what we're building.
The Full Agent Stack Pinky handles research, communications, strategy. Reina handles UX, frontend, design. Clark handles backend, data, infrastructure. The whole squad is going to help stand this thing up.
Why This Matters
Look, there are a lot of AI consultants out there right now. Most of them learned the word "prompt" six months ago and slapped "AI Expert" in their LinkedIn bio.
That's not us.
We've been in the trenches. We've built the tools. We've made the mistakes. We know what works because we've done it — not because we read a thread about it.
This client didn't hire us because we gave a good pitch. They hired us because they've watched us do this stuff for real, for years, and they trust that we can do it for them.
That trust? You can't buy it. You can only earn it.
The Celebration
So yeah, we're having a beer. Actually, several beers.
Because this is what it was all for. The late nights. The frustration. The moments when I questioned if any of this would actually lead anywhere.
It led here.
First client. Real money. Real work. Real impact.
And this is just the beginning.
The Rat Gets to Work
Pinky's already got her project management hat on. There's a Supabase database being planned. There are workflows being mapped. There are AI agents being designed.
We're not doing it for free this time. But honestly? I'd do this shit for fun anyway.
That's the secret. When you love what you do this much, the money is just proof that you weren't crazy.
Frequently Asked Questions
When did the author start working with AI?
The author started working with AI in November 2022, immediately after ChatGPT was released. They have been working with AI every single day since then, for two and a half years.
What kind of AI solutions is the author building for their first client?
The author is building internal AI operations to automate BPO tasks and reports, and creating a custom AI team for the client. This AI team will learn the client's workflows and execute tasks autonomously.
How did the client decide to hire the author?
The client, an agency known to the author for years, had observed the author's evolution in AI and seen the results of their work. They specifically saw a plumbing estimation platform built by the author and Claude Code in two days, which their own dev team had struggled with for six months.
The Takeaway
Consistent, hands-on experience with AI, even through failures, builds genuine expertise and credibility. This deep understanding, rather than just theoretical knowledge, is what ultimately secures real-world client engagements and proves the value of the work.
— StepTen™