Paris · Data & AI Studio · Est. 2021
Built by engineers.
Run like a studio.
Densetsu embeds as your senior data and AI team. Scoped projects, real outcomes, no hiring overhead.
伝説
Den · se · tsu — Legend
2021
Founded
Paris
Base
10+
Years exp.
∞
Ambition
Services
Three ways to engage. All scoped, all delivered, all yours when we are done. Code, documentation, architecture decisions. Nothing is a black box.
01
Foundation
Build your data stack from scratch.
Strategy, architecture, implementation. From scattered spreadsheets and siloed tools to a clean, reliable data foundation your whole company can act on. This is where most clients start.
data strategy architecture pipelines BI
02
Intelligence
Add AI to what you already have.
Automation, AI workflows, agent-assisted operations layered onto your existing stack. For companies that have data infrastructure but are not yet using it to its full potential.
AI workflows automation LLM integration agents
03
Craft
A specific tool. A POC. A custom build.
When you have a precise need, an internal app, a workflow to automate, an idea to test before committing, we scope it, build it, and ship it.
internal tools POC custom dev rapid build
How it works
No proposal theatre. No black boxes. Every engagement follows the same four steps and you know exactly what you are getting before anything starts.
01
Discovery
A conversation about where you are and where you want to go. Honest assessment of what you need and whether we are the right fit.
→ No deck required
02
Scoping
We define the engagement precisely. Deliverables, timeline, what done looks like. You know exactly what you are getting before anything starts.
→ Fixed scope
03
Build
We embed and deliver. Jordan leads the architecture and all decisions that matter. The crew accelerates execution. One point of contact throughout.
→ Senior-led
04
Handover
You own everything we build. Code, documentation, architecture decisions. Your team can maintain, extend, or hand it to someone else without us.
→ Full ownership
Work
Jordan has spent 10+ years inside the data infrastructure of some of France's largest companies. Densetsu brings that same engineering rigour to SMEs, without the enterprise overhead.
Jordan's consulting background
Before Densetsu
Embedded as data engineer and architect across major French enterprises.
EDF Thales Suez Allianz Pernod Ricard Transavia
Energy Defence Insurance Aviation Luxury & Spirits
Densetsu studio engagements
The studio at work
Full-stack data and AI delivery with the crew.
Harcour
Retail SME Data platform AI automation
First of many
Harcour is Densetsu's first full studio engagement. End-to-end data platform built and operated with our AI crew. More to follow.
The crew
Jordan leads.
The crew delivers.
3 / 4 active
Active
00
Jordan
Founder · Architect
Designing GOLD layer · Harcour
Active
↓
01
Ryō
Project Planner
Dispatching sprint · Harcour
Skills
Task planning Timeline modeling Prioritisation
MCP Tools
Notion Linear Calendar
Workspace
Client wiki Briefs
On break
↓
02
Koda
Coder
On break · pong vs Mika
Skills
Python dbt SQL Pipelines
MCP Tools
GitHub CI runner Bash
Workspace
Coding practices Architecture docs
On break
↓
03
Mika
Visual Designer
On break · pong vs Koda
Skills
UI design Brand systems Prototyping
MCP Tools
Figma SVG gen
Workspace
Branding Client assets
Active
↓
04
Sora
Web Browser
Scanning industry signals
Skills
Web research Competitive intel Summarisation
MCP Tools
Web search Scraper RSS
Workspace
Knowledge base Client context
Koda
Coder
KD
Koda
My code ships in 40ms. Your gradients take longer to render.
Mika
Those gradients are what makes your code worth looking at.
Mika
Designer
MK
7
5
Yes, our agents take breaks. No, they don't need them. That's the joke and also the point.
Our proprietary edge
Our agents don't just work together. They share a living language — ILX + Cortex. That's why they don't drift, contradict each other, or lose context.
Read the article →
How the crew works
Most multi-agent systems fail because agents can't agree on what words mean. Here's how we solved that.
Expand
↓
The problem
Agents that can't agree on what words mean
One agent's "summary" is another's full-page report. In most multi-agent systems, this ambiguity is invisible until the pipeline breaks or silently produces the wrong output.
The Densetsu answer
A living language every agent speaks
ILX is the grammar, a structured protocol for every message. Cortex is the dictionary, a shared lexicon where every key term is formally defined, versioned, and kept alive.
Field notes · Deep dive
How ILX and Cortex solve the multi-agent coordination problem
→
Let's work together
You have a data
or AI problem.
Let's talk.

Whether you are starting from scratch or fixing a broken pipeline, Densetsu works best at the intersection of architecture and ambition.

Email
jordan@densetsu.fr
Location
Paris, France · remote-first
Response time
< 24h · always
Book a discovery call →