I'm Nauren — Co-Founder and CEO of erxes, an open-source AI business operating system. Since February we've been running 20x Masterclasses, helping non-technical people build their own AI platforms from scratch.

15 cohorts. 5 days each. 2 hours a day.

When we started the 20x Masterclass in February, I wasn't sure non-technical people could actually build their own platforms, assistants, and websites from scratch.

I was wrong.

The ones who struggled least weren't the tech-savvy ones. They were the ones who knew their business problem so deeply that the tool almost built itself around them. After 500 graduates, here's what I actually learned.

Lesson 1: The people who grow fastest are the ones who believe it will work for them

Not the most technical. Not the most experienced.

The ones who showed up thinking — this is going to change how I work — and invested the time to find out.

Just 20 hours. That's all it took.

The people who protected those 20 hours came out at a completely different level. Not slightly better. Completely different.

We're all busy. Every single one of us has a reason not to start.

But the ones who made the room for it found a new way to work they can't imagine living without.

The technical barrier was never the real barrier. The real barrier was believing you were allowed to build it yourself.

You are.

Lesson 2: The real struggle isn't the technology. It's the clarity.

When you hire a new assistant in real life — what's the first thing you do?

You outline the work.

AI agents are no different. And that's where most people freeze — not because the technology is hard, but because they've never had to think about their own work this clearly before. Because our work is messy. The world is imperfect. Things happen in the middle that nobody planned for.

Start small.

When horses were the only transport, nobody could imagine a car. Not because cars were impossible — but because their minds were shaped by horses. Your mind is shaped by how you currently work.

So pick one task. The one that eats the most of your time. Draft it from start to finish — every step, every decision, every piece of context someone would need to do it without asking you questions.

Build your agent around that one task first. One task. Done properly. That's where it starts.

Lesson 3: Information is what separates a great AI agent from a glorified chatbot.

A typical AI chat model knows everything about the world in general and nothing about your business specifically.

An AI agent that actually works knows your business — your products, your processes, your customers, your language — and uses that knowledge to do real work.

The difference is information.

Ask yourself: what does your agent need to know to do its job without asking you every five minutes?

A few examples:

• Customer service agent — your product catalog, return policy, common complaints, tone of voice, how VIP customers should be treated.
• Sales agent — your pricing, pitch, objection responses, ideal customer profile, the exact words your best salespeople use.
• Operations agent — supplier contacts, reorder thresholds, approval chains, escalation rules, what counts as an emergency.

Now the real question: how does your agent access that information without you telling them every time?

This is where most people stop. They build the agent, give it a few instructions, and wonder why it keeps asking things it should already know.

The answer is a company brain — a living, organized knowledge base your agents treat as their source of truth. When your agent can find what it needs without you — that's when it stops being a tool and starts being a team member.

The bottom line

500 people walked into our Masterclass thinking AI was for other companies, other countries, bigger budgets.

Most of them left having built something real.

Start with belief. Start with one task. Start with the information your agent needs to work without you.

That's it. That's the whole thing.

The rest is just practice.