AI architecture for organisations that build things that last
Not a consultancy. Not a software vendor. One person who designs systems, deploys them, and makes sure they run without requiring permanent supervision.
The work starts with what you actually do — not with what AI can theoretically do.
What this concretely is
Autonomous workflows running in production. Systems that process information, make decisions within defined parameters, and pass results to the right person or tool — without a human in the loop for every step.
Examples of what has been built and is currently running:
- Prospection systems that qualify leads, draft personalised outreach, and log results — autonomously, daily
- Meeting pipelines that transcribe audio, generate summaries and quotes, and push to Drive and Sheets
- Monitoring dashboards that pull data from multiple sources and surface what needs attention
- Content systems that produce structured outputs from voice or text inputs
The tools used: n8n, Claude (Anthropic), Node.js, Google Workspace APIs, SMTP infrastructure. No proprietary lock-in.
Service packs
| Pack | Price | Duration | Deliverable |
|---|---|---|---|
| Discovery Call | 190 € | 90 min | Diagnostic of your current operations + written note identifying 3–5 automation opportunities with estimated effort and impact |
| AI Audit | 490 € | 5–10 days | Targeted analysis of one process or system + written implementation roadmap with tool recommendations and realistic timelines |
| Pilot | 2 900 € | 3–4 weeks | One workflow live in your environment + 2h hands-on training for your team |
| Full Integration | 9 500 € | 8–12 weeks | 3 to 5 interconnected workflows + full documentation + 6h training across two sessions |
| Retainer | 890 €/month | Ongoing | Monthly review call + email support + incremental evolutions as your needs change |
All prices are VAT-exempt under French micro-enterprise regime (art. 293 B CGI).
The Discovery Call is the logical starting point. It costs 190 €. If you decide to go further, that amount is deducted from the next pack.
For whom
SMEs and independent professionals
You have a process that takes too much time — qualification, follow-up, reporting, document processing. You want it handled without hiring someone to do it manually. Budget matters. Timeline matters. You need something that works, not a six-month project.
Cultural institutions and museums
You manage complex programmes, partners, press contacts, grant applications. Much of this is structured and repeatable. AI can handle the repeatable parts. The Discovery Call is a good way to find out which ones.
NGOs, foundations, and development organisations
Operating across multiple countries with limited central staff. Coordination, reporting, and communication workflows are often the bottleneck. The right system — one that runs autonomously and surfaces exceptions — changes what a small team can do.
Process
1. Discovery Call (90 min)
You describe your operations. I ask questions. At the end, you receive a written note — what could be automated, what shouldn't be, and what a realistic next step looks like.
2. Scoping
If we go further, I map the process in detail: inputs, outputs, decision points, edge cases, integration constraints. This becomes the spec.
3. Build and test
I build in your environment or a staging environment. You test with real data before anything goes live.
4. Handover and training
Documentation written for the people who will use and maintain the system. Training sessions focused on what can go wrong and how to fix it.
What makes this different
Most AI consultants come from one of two directions: pure technology (the tool is the answer) or pure strategy (the framework is the answer). Neither builds the thing.
This practice sits at an unusual intersection: digital artist and AI architect. Twenty years of designing systems — interactive installations, game interfaces, generative environments — that have to work reliably in public, often without any technical support on site.
That constraint shapes how I build AI systems. They are designed to fail gracefully, to surface problems clearly, and to run without requiring permanent human oversight. Not because that is a good philosophy. Because it is the only way a system remains useful six months after deployment.
Frequently asked questions
Do I need a technical team to work with you?
No. The systems are built to be operated by non-technical staff. Documentation is written for the people who will actually use it, not for developers.
What if the process I want to automate is more complex than it looks?
This is common. The AI Audit is designed to surface that complexity before any build starts. Better to discover it at 490 € than at 9 500 €.
Can you work with organisations outside France?
Yes. Current partners include organisations in Morocco. Remote work is standard. On-site work in Europe and North Africa is available for larger projects.
What tools do you use?
Primarily n8n (workflow orchestration), Claude (Anthropic) for language tasks, Node.js for custom logic, and standard cloud APIs (Google Workspace, SMTP, etc.). The choice of tools depends on what makes sense for your infrastructure — not on what I prefer.
How long does it take to see results?
The Pilot delivers one live workflow in 3–4 weeks. That is the fastest path from conversation to something running in production. The Discovery Call can happen within a week of first contact.
Contact
Email: [email protected]
Book a 30-minute call: calendly.com/kamelghabte/30min
SIRET: 489 325 308 00026 — Bordeaux, France