Skip to content
06 · Service 06 · AI · Automation

Automation & Artificial Intelligence

Make workflows, Claude-powered content production (product sheets, translations, reviews), custom bots, MCP integrations, repeatable playbooks.

05 outils
Make · Claude · MCP · n8n · GPT
12
Workflows delivered
AI-first
Systemic approach
Documenté
No black box
The problem

Where things usually get stuck.

  • Your team spends 15-25 hours a week on copy-paste: orders to CRM, invoices to accounting, parcel tracking to customer, sales to Google Sheets.
  • You have 800 product pages to translate into 4 languages and the agency quote is EUR 12,000 — you've been putting it off for 18 months.
  • You pay for 6 "AI-powered" SaaS at EUR 49/month each, all doing the same thing — and worse than a well-built Make workflow.
  • Nobody on your team can write a serious prompt, so ChatGPT gets used like Google and the outputs are mediocre.
  • When an automation breaks (API change, expired token, unexpected format), nobody knows how to fix it and everything stops.
My solution

The approach I bring you.

  • 5-day "lost time" audit: I shadow your ops, time tasks, deliver a priority list ranked by hourly ROI — every automation costed in hours saved per month.
  • Make or n8n workflows built, tested and documented in 2 to 4 weeks depending on complexity — with error handling, retries, Slack/email alerts when it breaks.
  • Content production at scale via Claude Code: 800 product pages FR/DE/NL/IT/ES in 10 days, human-validated by sample, marginal cost crushed.
  • Specialist AI agents via MCP (Model Context Protocol) — not a chatbot, but an agent that connects to your tools (WooCommerce, GSC, Google Ads, BigQuery) and executes real tasks.
  • Documented playbook + Git-versioned code: your workflows survive my departure, your team can modify them without calling me.
How it unfolds

Four steps, documented.

No black box: every phase has a concrete deliverable and a time window agreed up front. You know exactly where we stand, at any moment.

01 Audit

"Wasted time" audit

Cartographie des tâches répétitives, des copier-coller entre outils et des copies manuelles. On chiffre l'ROI en heures par mois avant de coder.

02 Plan

Workflow architecture

Choix Make vs n8n vs Apps Script, sources de vérité, points d'intégration MCP, modèle de coûts API. Tout sur papier avant le clic.

03 Build

Construction & branchements

Make workflows, Claude agents via MCP, WooCommerce / Sheets / CRM / marketplace integrations. Versioned, commented, transferable.

04 Iterate

Documentation & handover

Written playbooks, smart alerts, API cost oversight and training for your team. No black box.

In concrete terms

What you get, line by line.

01 On-site ops audit (1 week): map of repetitive tasks, timing, hourly-ROI scoring, prioritised automation plan
02 Production-ready Make or n8n workflows: WooCommerce <-> Google Sheets <-> CRM (HubSpot/Pipedrive) <-> Gmail <-> Sendinblue <-> Amazon Seller Central
03 Multilingual product page generation via the Claude API: Sheets -> Claude -> WooCommerce REST pipeline, sample-based human QC
04 Catalogue translation at scale: 500 to 5,000 listings FR <-> DE/NL/IT/ES/EN with SEO preserved and cultural adaptation
05 Custom Claude Code agents: plugged into your stack via MCP, capable of auditing GSC, writing SEO briefs, pushing Merchant Center feed updates
06 Automated multi-source reporting: Google Ads + Meta Ads + GA4 + GSC + WooCommerce -> BigQuery -> Looker Studio, refreshed daily
07 Smart Slack/email alerts: ROAS drop > 30 percent, sales anomaly, stock-out, SEO top-10 loss, Merchant Center error
08 Multi-channel stock/price sync: WooCommerce <-> Amazon <-> Etsy <-> Cdiscount <-> Manomano, with priority rules and conflict handling
09 Operational AI chatbots: level-1 support, lead qualification, product FAQ, plugged into your real catalogue via RAG
10 Python or Apps Script for heavy lifts: product-feed clean-up, CRM dedup, contact enrichment via Clearbit/Apollo
11 Documented playbook (Notion or Markdown in Git): workflow diagram, failure points, repair runbook, API contacts
12 Team training over 2 half-days: how to read a Make workflow, how to debug, how to add a simple module
Tools & technology

The toolkit I use day in, day out.

Claude (Anthropic) Claude Code Claude API MCP (Model Context Protocol) Make.com n8n Zapier Google Apps Script Python BigQuery Looker Studio OpenAI API Pinecone (RAG) GitHub Actions
AI & automation capabilities

Workflows documented, pas magiques.

What's actually under the bonnet. Each column is a working sub-system, not a marketing line.

01 Capability

Make & n8n workflows

  • Branchements WooCommerce ↔ Sheets ↔ CRM ↔ Gmail ↔ marketplaces.
  • Reporting Google Ads / Meta / GA4 automatisé vers Sheets ou Looker.
  • Alertes intelligentes : ROAS, rupture stock, chute de position, anomalies.
Make · n8n selon volume & coût mensuel
02 Capability

Claude agents via MCP

  • Serveurs MCP sur mesure pour vos outils internes et bases de données.
  • Agents spécialisés : fiches produits, traductions, audits, reporting.
  • Claude Code orchestré pour les opérations quotidiennes répétitives.
MCP natif peu maîtrisé en France
03 Capability

AI content production

  • Fiches produits dans 5 langues avec validation humaine native.
  • Traductions cohérentes pilotées par glossaires et style guides.
  • Articles SEO et GEO avec citations sourcées, pas de hallucination.
5 langues FR · DE · NL · IT · ES
Why me

What sets me apart in concrete terms.

01

I sell systems I run myself — 18 WordPress sites synced by a plugin pair I wrote, a Claude Code worker on a nightly cron, an Instagram automation pipeline in Python + Sheets.

02

Real Claude Code + MCP fluency — Still rare in the French-speaking market — not picked up from a 2-hour YouTube course.

03

10 years of e-commerce + paid media — I know which processes are worth automating and which are micro-optimisations with no ROI.

04

Code versionné Git, documenté, sans dépendance à ma personne — pas de boîte noire « contactez-moi pour modifier », c'est votre système

The real questions

What founders tell me ask me before you sign.

Not the generic "why choose us" FAQ — the real questions about budget, timeline and scope you're asking yourself at 10pm.

A more specific question?
  • What kind of companies do you work with?

    Mostly e-commerce businesses with €500K to €15M in turnover, plus agencies and multi-site SMEs that need to industrialise repetitive work. I rarely work with large corporates — my value sits in agility, not in committee management.

  • What's the typical ROI for a Make workflow?

    A well-built workflow saves 5 to 30 hours a month for a Make subscription cost of €20 to €100 per month. ROI is typically paid back in 2 to 4 months. I quantify the hours saved before writing any code, not after.

  • What exactly is MCP?

    Model Context Protocol — an open standard released by Anthropic in 2024 that lets Claude (or any compatible LLM) connect securely to your internal tools. In practice, your WooCommerce catalogue, your Sheets or your CRM become accessible to an AI agent through a dedicated MCP server.

Let's start

Ready to take action on automation & artificial intelligence?

Let's discuss your project