We build critical software that runs your business.

Who we are

code.store was built on a simple idea: custom software is becoming a commodity.

After founding Adyax, a 500-person Drupal CMS leader sold to Smile Group in 2018, we set out to build Europe's most efficient software factory.

Today: 80 experts, headquartered in Paris, serving clients worldwide.

France United Kingdom United States Brazil United Arab Emirates South Africa Colombia Ireland Italy Germany Oman
code.store team
team / 2026
How we build

We redesigned every aspect of software production.

01

Teams

Stable, senior, top 1% business-oriented developers — supported by UX profiles who can ship React front-ends and data experts involved from day one.

02

New Methodology

Prototypes as specs, weekly deliveries, full-scope automated test coverage, and night-shift AI agents that refactor and debug continuously.

03

New Standards

Infrastructure description, CI/CD pipelines, reusable frameworks, and shared engineering conventions.

04

Tooling

RETRO for legacy software retro-documentation, Usedge for UX research at scale with AI, and hundreds of Claude Code skills and Markdown instructions to scale AI-assisted delivery.

code.store team at work
the factory at work
What becomes possible

With our software factory — embedded engineers, tools and AI agents — we can build things with your teams that were not possible before

WHAT WE DO? 1 / 4

Legacy software migration & rebuild

5 millions of COBOL and no-one really knows how it works or where the application is being used? Modernization of such software was a typical project no IT director would like to start… before.

We migrate and rebuild large legacy systems end-to-end on modern stacks.

COBOL, PL/SQL, DB2, NSDK, Natural, AS/400, old Java or PHP: we retro-document the existing software, extract business rules and user stories, then rebuild the application from the ground up.

We modernize the UI, expose APIs, replace obsolete code with ETLs and BI products where relevant, and secure the full migration with automated end-to-end tests.

We also map, clean, and migrate your data.

RETRO — AI retro-documentation tool
COBOL
COBOL
DB2
DB2
PL/SQL
PL/SQL
Natural
NATURAL
retro · extract · rebuild
React Node.js / Java Postgres APIs E2E tests Kubernetes Clean UX AI
WHAT WE DO? 1 / 4 EXAMPLE OF LEGACY MIGRATION PROJECT…

Migrating AS/400 ERP to a modern Node.js / React stack

Picto is an iconic company and a global expert in analog photo printing, working with most French museums, as well as LVMH and Richemont. Their aging IBM AS/400 ERP system was replaced with a custom AI-first build, using RETRO to extract all business rules and the data model.

Key figures
250employees
100kinvoices
500providers
4countries
45kprojects
8 months · 300K€
WHAT WE DO? 2 / 4

Your sovereign software instead of SaaS

CRMs and ERPs are expensive — and rarely fit your business perfectly.

We build custom software designed around the way your company actually works: co-designed with your users, integrated into your IT ecosystem, and deployed in your own cloud.

We can run it for you, or transfer full ownership to your teams. If you have developers, we bring them into the project, train them on our processes, and equip them with our tools.

Your software. Your data. Your cloud. Up to 10× cheaper to run than SaaS.

Example
WHAT WE DO? 2 / 4 EXAMPLE SAAS REPLACEMENT

Worldwide Medical Assistance management system

Replacing Salesforce, Microsoft Dynamics and 10 legacy systems across 30 countries — a 24/7 critical platform handling personal medical data and payments to hospitals and providers around the globe. Delivered in 9 months for 1M€. Annual run cost dropped from 2M€ to 200K€.

Key figures
1,500agents
30countries
200M€pay-outs
75kcases / year
5portals
9 months · 1M€
WHAT WE DO? 3 / 4

Complex data mapping & migration at scale

Every major software transformation becomes a data migration challenge.

Moving from one platform to another, from custom software to SaaS, or from SaaS back to custom means dealing with different data models, objects, attributes, relationships, and storage logic.

Migrating large, heterogeneous datasets is hard — and usually takes far more time than expected.

We have built deep expertise in AI-assisted data migration: mapping, cleaning, transforming, and migrating your data from one source to another, with a 100% recovery guarantee.

The replayable data migration pipeline — schema
WHAT WE DO? 3 / 4 EXAMPLE OF CMS MIGRATION

Migrating 2 million assets to a modern CMS

L'Express is a weekly print and digital magazine, a major French brand. Their outdated home-brew CMS — built on Symfony 1.0 — needed 4 full-time back-end developers just for maintenance, leaving product teams stuck with no new features for two years. We migrated them to Arc XP and rebuilt their front-end and mobile apps from the ground up.

Key figures
100journalists
80templates
40+API endpoints
2Mmigrated assets
800kredirects
6 months · 300K€
WHAT WE DO? 4 / 4

Automation of processes on top of your tools

Most companies run key processes across several old, heterogeneous applications that are hard to change.

People still analyse supplier quotes manually, write the same emails again and again, copy data from PDFs into ERPs, update old CRMs from emails, or rebuild sales reports from calendars, inboxes and CRM data.

We automate these workflows with lightweight agents that combine deterministic rules, LLMs, API calls and UI actions.

They don't require a full system replacement: they sit on top of your existing tools, simulate human actions when needed, and make messy operational work faster, safer and more reliable.

Workflows we automate
  • Supplier-quote analysis
  • Repetitive email drafting
  • PDF → ERP data entry
  • CRM updates from inboxes
  • Sales reports from mail & calendars
No system replacement Sits on top of your stack
How we work

Let's look at exactly how we work with your teams to make it happen — and where AI helps us move faster than ever.

_
HOW WE WORK 01

Research first, then a working prototype

First, we run large-scale UX research sessions with your users, powered by AI, using the Usedge platform. This gives us sharper insights, a better understanding of real needs, and stronger UX decisions.

Then, instead of spending weeks on static Figma screens, we immediately generate a fully functional React prototype — dramatically reducing the time to start the project and cutting down back-and-forth iterations during design.

UX research and prototyping workflow — schema
HOW WE WORK 02

Agentic + human · 1-week sprint

Our agentic sprint treats verification as the binding constraint: an executable six-part spec gates every build, work-in-progress is capped at review capacity, and we demo only what's verified — never what's merely generated.

Agents generate in parallel under deterministic phase gates and one-file-one-owner isolation. Humans hold design review on the risky surfaces — schema, auth, performance, contracts — and every agent failure feeds back into the shared constitution, so quality compounds sprint over sprint.

Agentic + human sprint — build pipeline schema
HOW WE WORK 03 LIVE PROTOTYPE

Example of a designed prototype

A small example of a simple project: a mobile application for a French life-insurance provider. It offers every feature to manage life-insurance contracts and track their valorisation.

Fully functional React prototype — interact with it →

code.store founders

Better call code.store!

We'd be more than happy to help you and your team.

From Paris, with love.