African companies deserve to know what the world says about them — in real time, in their language, on their phone.
Moroccan and African companies are talked about daily in media and AI engines — but they have no way to know. Existing tools are English-only, US-focused, and built for Twitter.
A reputation intelligence pipeline built for francophone and African reality. Arabic sources, French business press, the AI engines your customers actually use. Delivered on WhatsApp.
"We believe reputation should not be a black box. Every score we deliver can be traced to a source article, an AI response, and a sentiment classification with its confidence. We build in public because that's how we'd want to be treated as a client."
Built by one person, for now.
Harch Atelier is founded and operated by a single person — a software engineer and analyst based in Casablanca. No VC, no cofounder drama, no inflated team. Just one person obsessed with reputation data.
Why I started Harch Atelier
In late 2023, I was helping a Moroccan bank understand why their customer satisfaction scores were dropping. The answer wasn't in their internal data — it was in what Hespress and L'Économiste were saying about a new fee policy. But nobody at the bank had noticed for three weeks.
I looked for a tool that could have warned them. The francophone options were press clipping services from the 1990s. The modern options (Meltwater, Brandwatch) were English-only, priced for Fortune 500s, and didn't track what ChatGPT says about you.
So I built one. The first version was a Python script crawling five media sources and querying ChatGPT every hour. It ran on a single brand for 30 days. The data was so striking that the brand became our first paying client.
Today we track 30+ media sources, 8 AI engines, and 3 languages. We have paying clients in banking, telecom, energy, and hospitality. We're still small, still founder-operated, and still building in public.
What's under the hood.
We publish our full tech stack because we believe in building in public. Four layers — collecte, ingestion, analyse, livraison — each with battle-tested open-source and frontier LLM components.
The story so far.
We publish our milestones — the wins and the pivots. Here is the full timeline from idea to today.
The idea
Founder notices Moroccan companies have no way to track what AI engines say about them. Existing tools are English-only, US-focused.
First prototype
A Python script crawling 5 Moroccan media + querying ChatGPT. Runs on a single brand for 30 days. The data is striking.
First paying client
A Casablanca bank signs up for monthly monitoring. The PDF report is 12 pages. They renew.
Harch Atelier launches
The product becomes a subsidiary of Harch Corp. Three tiers, transparent pricing, bank transfer only.
10 clients · 4 sectors
Banking, telecom, energy, hospitality. The HARCH 100 ranking is born — a public benchmark of Moroccan corporate reputation.
8 AI engines tracked
We add Gemini, Claude, Copilot, Mistral, and Grok to the original three (ChatGPT, Perplexity, Google AI Overviews).
Building in public
We publish our methodology, our pricing, our tech stack. We share what we learn about AI reputation in francophone markets.
Harch Atelier in 2025.
Real numbers, updated quarterly. No vanity metrics, no inflated figures.
Four things we won't compromise on.
These aren't marketing copy. They're the principles we use to make product decisions, pricing decisions, and hiring decisions.
Building in public
We publish our methodology, our pricing, our tech stack. No black box, no magic score. You see exactly how we work.
Francophone first
Built for Morocco, Africa, and the francophone world. Arabic sources, French business press, Darija detection.
No lock-in
Monthly contracts. Bank transfer. Export your data anytime. If we're not delivering value, you should leave.
Real humans
No chatbot-only support. You get a real analyst who knows your brand, your sector, and your reputation history.