MONA's own AI story · MONA 2.0

A real renewal · not a single made-up number

Why the 20 of April?

On 16 April, the model Claude 4.7 launched. Exactly 4 days later, MONA decided to renew the entire way a company of more than 200 people operates, with AI. Looking at the 4 days, many people think it was an impulsive decision. In fact, we had been patiently waiting for exactly this moment for many years.

16/4

Claude 4.7 launches: the threshold appears for the first time

4 days

from the threshold to the decision to renew everything

60 days

of continuous renewal of how the company operates

250

staff, turned into our own laboratory

🗓️ A moment we had waited years for

Not impulsive. It was the day the technology became just mature enough to hand whole jobs to machines

For many years, AI was good at pieces but not good enough to hand over a WHOLE stage. It answered well, wrote well, but still needed a person in the middle: gathering data, connecting tools, fixing errors, putting it back together. That is not automation; that is an analysis machine sitting idle, waiting for someone to feed it.

On 16/4, that threshold broke. Not because of a smarter chatbot, but because for the first time all three pieces were in place for an AI to take on an entire process from start to finish by itself:

🧠

Deep enough reasoning

Model 4.7 doesn't just answer; it plans multiple steps, corrects itself along the way, and keeps the thread of a long task without getting lost. Enough to TRUST it with a real stage of work, not just a single question.

⌨️

Claude Code: knows how to do the work

The agent reads, writes and runs commands on real systems by itself. It doesn't stop at suggestions; it can build software, fix processes and assemble tools. The hands and feet of the brain.

🤝

Cowork: knows how to coordinate

Several agents take on one job together, using real tools like a staff member: looking things up, operating, handing off to each other. For the first time AI doesn't just chat; it DOES the work, and does it together.

Put those three pieces together: for the first time an AI takes on an end-to-end stage of work without human hands patching it in the middle. Only then could we use the word automation.

🔘 Press the button, and don't turn back

MONA only sells what we actually use ourselves

Once we saw clearly that the threshold had arrived, we didn't rush to sell it to clients first — MONA used it on ourselves first. The result: every single number came out oddly good.

As a company of 250 people, our monthly organic leads — with no ads and not a single marketing campaign — already exceed 4000+, we weren't hesitant or afraid of change, afraid of "breaking something". Quite the opposite: MONA is one of the few businesses that has both deep technical know-how and a "huge prize" waiting if automation succeeds.

So we rolled it out at a pace many people called reckless: fast, decisive, with no back door left open. Because half-hearted automation is the most expensive kind — doing it for show and dropping it halfway costs you money and your own team's trust.

⚠️ Misconception number one

Handing out AI accounts to the whole company is not automation

Most businesses start their AI transformation with one move: buy ChatGPT, Gemini, NotebookLM, Canva... then hand out accounts to staff and say "everyone use AI to be more productive". It sounds very reasonable. But it almost always fails, and it fails at three levels:

Light

Break-even — nothing changes. A few curious people use it for a few days and then forget. The tools sit there, the process stays exactly the same. The license money quietly evaporates and nobody notices.

Serious

It produces junk. People without enough skill use AI to rush through the part they should have thought about — clichéd content, wrong reports, shallow work multiplied many times over. Productivity creates junk, not value.

Very serious

A resistance movement. Staff stuck in a 2005 mindset see AI as a threat to their jobs. They mock it, push back, refuse, form camps — turning a good initiative into an internal battlefield.

Automation is not handing out tools. It is redesigning how a step of the work runs on its own — so machines carry the repetitive part and people are free to do the valuable part.

🚀 A golden moment for those who move first

Mature technology is the biggest gift to businesses that already have a foundation

There's a misconception that AI exists so that young people with nothing can overtake those who came before. In fact it's the opposite. The hardest part of applying AI in a business is knowing which tasks are worth automating, which processes are core, and what the quality standard is — things you only have once you've been in the trade long enough to no longer be a rookie.

When a new technology launches, the people who benefit most aren't the best at technology — they're the ones with a solid foundation who are quick enough to jump in early. You already have clients, processes, a team and standards. AI is just a lever placed on that existing foundation.

A business with a solid foundation that also moves quickly: ahead catches up with the big players who started decades earlier; sau opens a gap that newcomers to the trade can't close. This window won't stay open forever.

🧩 MONA's core mindset

Fix from the top down, accept the second-best solution — as long as it's the right people at the right time

This is the core software mindset at MONA that many clients and partners don't yet understand. Nobody starts with a perfect blueprint covering the whole company. We do it the other way round: fix from the top, start from a specific pain point, and work down to the roots.

Pick the most painful spot

Start with the most bureaucratic, repetitive, people-hungry step — not the easiest one. Solve one real pain point for good.

Let the front line taste the results first

Let the very people doing the work by hand see an AI Agent genuinely take work off their plate. When they see with their own eyes that it works, trust spreads on its own — stronger than any presentation.

Go deeper into the roots — with the very people who resisted

Once it's clear that it works, roll it out deeper into the roots. The surprise: the people who used to resist now back it, because they've seen AI lift them up rather than replace them.

Systematise it into one all-covering OS

Only at the end do we assemble everything into one large, closed-loop machine. Even if the software has to be redesigned from scratch — because there was no big plan at the start — that's a price that has to be paid, and is worth paying.

We aren't obsessed with perfection. A second- or third-best solution that fits the business best at that moment always beats a perfect-on-paper solution that never gets finished.

📚 Further reading