The Operator’s Advantage: AI as a Margin Lever for Small Business

A conversation with Mael Garbe — 19-year-old founder of ORBITECH and author of LBO Journal — on why the largest untapped opportunity in AI sits inside the unglamorous back-office workflows of traditional small and medium businesses, and why a founder’s edge comes from pairing engineering with the economics of ownership.

By Louise Servoin · 2026-06-17 · 9 min read

## Mael Garbe – Founder of ORBITECH and Author of LBO Journal

Focus: AI for Small Business, SME Modernization, Operator-Led Buyouts, Sovereign Cloud. Based in Lille, France.

Most of the noise around AI is about the frontier — bigger models, flashier consumer tools, the race to replace knowledge work. Mael Garbe is interested in something far less glamorous and, he argues, far more valuable: the millions of small and medium businesses still running on manual workflows, fragmented spreadsheets, and paper.

At nineteen, and based in the Hauts-de-France region, Mael is the founder of ORBITECH and the author of LBO Journal, an editorial project dissecting the financial mechanics of small-business buyouts. He sits at an unusual intersection — he writes production software and he studies how traditional companies are acquired and modernized — and from that vantage point, his read on AI is refreshingly unsentimental: it is not a product to sell, it is a lever that makes real businesses measurably more efficient.

We sat down with Mael to talk about where AI actually creates value for small business, why the back office beats the frontier, how he structured ORBITECH as a company, and what no algorithm will ever do for an operator.

## Where the Opportunity Actually Is

Most of the AI conversation is about flashy consumer tools and frontier models. As a founder, where do you actually see the biggest opportunity?

The opportunity isn’t the frontier — it’s the backlog. A huge share of traditional B2B service companies in regional France are run by owners approaching retirement, and these operations are weighed down by severe technical debt: outdated, manual IT workflows, fragmented Excel files, paper-based tracking. Everyone is staring at the cutting edge while the largest, most immediate gains are sitting in back offices nobody wants to talk about. When you automate that administrative drag, the margin you unlock is immediate and measurable. That’s the asymmetry — low glamour, high return.

"The opportunity isn’t the frontier — it’s the backlog. When you automate the administrative drag, the margin you unlock is immediate and measurable." — Mael Garbe

## The Operator’s Math

You write LBO Journal and you also build the software. How do those two worlds combine into an actual business strategy?

They’re deeply synergistic — that’s why I think of the path as a “Searcher-Developer.” If you approach a small-business acquisition purely as a financial investor, you look at leverage ratios — capping senior debt relative to EBITDA — and you calculate conservative Debt Service Coverage Ratios so the cash flows can cover the liabilities. That’s the finance half.

But if you also have an engineering background, you can audit the company’s workflows as a system of data pipelines. The moment you acquire the business, you step in and automate the administrative backlog, rewrite the database relational mapping, and implement lightweight custom tools. By modernizing the software layer, you instantly optimize the operating margin and accelerate the rate at which the business pays down its debt. Writing code and structuring a leveraged buyout are both exercises in designing self-sustaining, optimized machines — and AI collapses the cost of doing that modernization.

## Building the Company

ORBITECH is actually two legal entities — a commercial micro-enterprise and a non-profit association. Why design it that way as a founder?

As a founder, you design the container before you pour in the work. ORBITECH brings together two distinct legal entities with two complementary missions — deliberately kept separate, both legally and operationally.

The micro-enterprise is the commercial engine: it builds the AI ecosystem and the product for the French education market — the tools, the platform, the school-facing offer. The non-profit, a loi 1901 association currently being deployed, carries a different mission: free access to knowledge, with no financial barrier, for the widest possible audience. Keeping them separate protects both — the commercial entity can move at market speed, and the association keeps its mission clean. The product itself is built on the Socratic method: AI that guides reasoning step by step instead of handing over answers. But structurally, the discipline is the same one you’d apply to any company — decide what each entity is for, and don’t let the two blur.

## Sovereignty as a Business Decision

You host your entire stack on sovereign European infrastructure. For a small company, isn’t that slower and more expensive than just routing everything globally?

It’s a deliberate trade-off: sovereignty over comfort. When you handle educational data, especially for minors, compliance is not a marketing checkbox; it is a structural necessity. Concretely, 100% of our user data and backend run in France—on AWS in Paris (eu-west-3), via Supabase—under a strict GDPR framework, with Data Processing Agreements (DPAs) and Standard Contractual Clauses (SCCs), and with no user data ever used to train models. For inference, we use the Anthropic API today: it is the most technically viable option at this stage. The important point is that this is American compute power, not a transfer of data or of sovereignty. In parallel, a French inference solution is currently being evaluated.

It is far easier to route API calls globally without thinking about where the data lands, but for a business serving schools and SMEs, that architectural integrity becomes a competitive differentiator and a trust signal. It’s part of what allowed the project to be covered by regional media — Ici Nord and France 3 Nord reported on it in March 2026 — and to begin piloting the platform in private secondary schools in northern France.

## What No Algorithm Will Do

As AI commoditizes standard code and analysis, what’s the one capacity it can’t replace for an operator?

The algorithm can calculate financial ratios or write PostgreSQL schemas, but it cannot navigate human risk tolerance or empathy-driven negotiation. It cannot sit in a room with a retiring owner who has spent thirty years building their local SME, understand their personal anxieties about handing over their legacy, and build the deep personal trust required to close a deal. AI is a co-pilot that reduces cognitive load — but the destination must stay human-led. The ultimate measure of a career is never an automated promise; it’s the physical, operational trace of what you had the courage and discipline to actually build.

## The Founder’s Thesis

If you had to compress it to a single thesis for founders looking at AI right now, what would it be?

The edge isn’t in building the next model — it’s in being the operator who uses AI to make unglamorous, real businesses dramatically more efficient. A well-guided, Socratic use of AI compresses the path to comprehension and execution without short-circuiting it; the speed is real, but it serves building, not avoidance. Velocity in the service of autonomy, never as a substitute for it. The future belongs to people who pair AI fluency with the discipline to actually run and modernize a business.

Tags: AI for Small Business, SME Modernization, Operator-Led Buyouts, LBO, M&A, Sovereign Cloud, EdTech, Lille, Entrepreneurship