Customer story · ConstructionBatibig logo

Rebuilding a construction group's core system in four months, with Claude.

Batibig runs a fast-growing construction group. Stratos rebuilt the core of its business software, planning, client management and invoicing, in four months, using Claude across design, specification, code, testing and documentation.

2025 · Construction · Product rebuild, AI-accelerated delivery

4 months
To rebuild the core system, end to end
3
Core modules rebuilt: planning, client management, invoicing
The challenge

Modern management on an old-school industry, held back by old software.

Batibig was founded by two operators who bought a first construction company and brought modern management to a traditional trade, then rolled up more companies into a group now worth around €350M and heading toward €1B in revenue.

The group had outgrown its software. The system at the heart of the business, planning, client management and invoicing, still ran on a very old platform that no longer matched how the company worked. Batibig wanted to modernize that core, without a multi-year replacement project.

Batibig founders photo
Batibig group site photo
Build a custom, business-specific ERP capable of:
  • Standardizing management processes across all group subsidiaries (excluding public contracts)
  • Replacing three disparate tools (Everwin, React, Retool) with a single coherent platform
  • Interfacing with the existing ecosystem (invoicing via Pennylane, payroll via Silae, leave management via Lucca, approval workflows via Approval)
  • Handling complex, non-trivial building-maintenance-sector business rules: sector-specific VAT and reverse-charge VAT (auto-liquidation), contract amendments (avenants), accounting cut-off, managed properties
  • Meeting a tight delivery schedule: 4 months, from a Month 1 kickoff, with two intermediate functional milestones (Month 3, Month 4)
Our approach

Rebuild the core in four months, with Claude across the lifecycle.

Rather than a long specification-then-build cycle, we ran the project with Claude at each stage, from validating concepts with the client to writing the code, the tests and the documentation. The result was a full rebuild of the core system in four months.

How we did it with Claude

Claude across design, specification and build.

Three stages, each accelerated by Claude, with senior engineers keeping quality and architecture under control.

STEP 01

Design and concept validation

We created the mockups and the design system with Claude Design, so we could validate concepts, designs and ideas with the client quickly. Putting real interfaces in front of Batibig early meant decisions were made on something concrete, not on documents.

STEP 02

Specifications from the mockups

Once the mockups were validated, we had Claude write the specifications directly from them. The specs stayed tied to what the client had already approved, instead of drifting into a separate document.

STEP 03

Build, test and document

Senior developers set up the project on solid engineering practices, then we built with Claude Code on the Max plan, using a set of Skills, some from the market and some we built ourselves, fed by the developer specs prepared earlier. End-to-end tests and demo videos of the application were produced with Claude and Playwright, and reused, for example, to build the FAQ. All the technical documentation and the FAQ were written with Claude.

Batibig application screenshot
What's next

From documentation to a self-service assistant.

Because the full technical documentation and the FAQ were produced with Claude, they form a clean base for the next step: an assistant grounded in the complete documentation and FAQ, so Batibig's teams can get answers directly instead of searching.

Planned

A documentation-grounded chatbot

An assistant that answers from the complete documentation and FAQ built during the project, giving the client's teams a single, reliable place to ask.

Results

Why this wouldn't have been possible without Claude Code.

A custom ERP with this level of sector-specific depth (building maintenance, multi-subsidiary, complex tax rules) normally forces a trade-off between three variables: functional scope, quality of business fit, and timeline. Compressing one of the three to hold the other two is the norm in classic ERP integration.

Here, the production velocity enabled by Claude Code meant that trade-off didn't have to be made: scope stayed broad (8 modules, not a stripped-down MVP with 2-3 functions), business fit stayed precise (real sector-specific rules, not generic approximations), and the timeline held (4 months, both release milestones met).

Core system rebuildClaude DesignClaude CodeAI-accelerated deliveryTechnical documentation

Modernizing a core system?

Let's talk