We train companies on AI, and we build it into production. Two crafts, one foundation: measured proof.
📍 Pertuis, France · Agency since 2010
BULDEE builds AI systems that ship to production, and trains teams to run them without depending on us.
Three formats, one shared foundation: baseline measured on-site, use cases handled end to end, gain re-measured at D+15.
| Format | Duration | Price (excl. tax) |
|---|---|---|
| Entry format | 1 day · 7 h | €2,000 |
| Team workshops | 3 days · 21 h | €5,400 |
| Rollout + IT governance | 5 days · 35 h | €8,500 |
An AI squad and product scoped by an architect: scope, milestones and price fixed at kickoff, never staff-augmentation.
| Client | Mission | Duration / team size |
|---|---|---|
| Edenred | Fractional CTO, technical governance, architecture migration | 3 years · 40 developers |
| France Télévisions | Architecture rebuild, high traffic peaks | 2 years · 15 developers |
| Axialys | Drupal, contributions to Drupal Media core, training | 4 years · 8 developers |
Also worked with: GRDF, IAD, Le Figaro, Cours Florent.
| Product | What it is |
|---|---|
| Metrikia | Multi-touch attribution SaaS (Meta, Google, TikTok, Pinterest ↔ Stripe and 11 CRMs), 10 models including Markov and Shapley, 23-tool AI agent, hybrid pgvector RAG, MCP OAuth |
| Qualia Hub | AI-driven appointment scheduling on Instagram and WhatsApp, custom RAG |
| TheLab.io | E-learning and marketplace for independent coaches, Clean Architecture, Mercure/FrankenPHP |
| AI Craftsman Superpowers | Open-source craftsman ecosystem for AI agents (Claude Code, Hermes Agent, ...): architecture quality gates, refactoring, legacy rescue, testing, security review, AI/ML system design (RAG, agents) |
| SportHabits | Offline-first Flutter app, AI extraction of coaching programs |
We're an AI Studio first: every product ships an agent, a RAG pipeline, or both, in production, not a proof of concept.
AI & ML · what we actually build
Multi-tool agents, tool calling, reranking, evals: in production, not a demo.
Product engineering
Quality & infra
| Repository | Description |
|---|---|
ai-craftsman-superpowers |
Craftsman ecosystem for AI agents (Claude Code, Hermes Agent, ...): quality gates, refactoring, legacy rescue, testing, security, AI/ML system design |
galaxia-ads |
Claude Code plugin for ad analysis and operations, built on Metrikia |
metrikia-hermes |
Hermes Agent skills for Metrikia: weekly ROAS report, campaign audit, budget alert |
notebooklm-galaxia |
Claude Code plugin: NotebookLM access from the Galaxia stack |
claude-system-doctor |
Scheduled audit and self-heal for a Claude Code installation |
oh-my-privacy |
Blocks secrets before they reach an AI agent's context |
github-profile-optimizer |
Claude Code plugin to generate GitHub profile READMEs |
💡 Our client projects and internal products are private - Metrikia, TheLab.io, SportHabits, Qualia Hub. Contact us for a demo.
Alexandre Mallet, software architect and AI engineer, 15 years of experience. BULDEE founded in 2010, Drupal Media Core Contributor (2016-2019), 2 years at France Télévisions, 3 years as technical director at Edenred.
BULDEE is registered as a training provider under number 93840574684 (DREETS PACA). This registration does not constitute State approval. Qualiopi certification in progress.
A business process to automate, a team to train on AI?
📍 Pertuis, France · 🌐 buldee.com · 📧 contact@buldee.com
Technical excellence · Informed decisions · Measurable results