From Complex Workforce Planning to an AI-Assisted Shift Planner
A three-person Abylon team, including one developer working with Claude Code, delivered a working Shift Planner MVP in just 31 days. Over the following three months, Abylon and MOL Group’s Digital Factory continued to extend and refine the solution, combining AI-assisted development with project-defined engineering rules, automated testing, independent AI review and human approval.
About The Client
MOL Group’s Digital Factory is the innovation hub driving the digital transformation of the company’s Consumer Services business, with a focus on digitalizing customer interactions and internal operations. Its multidisciplinary team brings together product, engineering, analytics and design expertise to develop digital solutions across MOL Group.
Challenge
Managing workforce planning across approximately 2,400 MOL Group service stations in more than 10 countries comes with a level of complexity that spreadsheets and local processes were never designed to handle.
Different stations had developed different ways of working. Some teams relied on spreadsheets, others on paper-based processes, and scheduling practices varied between locations. This made it difficult to create a consistent view of workforce planning, coordinate schedules and keep track of changes.
At station level, managers were constantly balancing several moving parts: employee availability, holidays, shift coverage and daily workload targets. Even small changes could mean adjusting several elements of the schedule.
MOL Group’s Digital Factory therefore needed more than a digital version of an existing spreadsheet. It needed a central workforce planning platform that could connect scheduling with its existing Host Optimizer Tool, provide station-level access for partners and managers, and support a phased rollout across multiple countries.
The solution also needed to fit into MOL’s existing technology environment, support several user roles and languages, and remain flexible enough to evolve as new requirements emerged.
Solution
- Building a Shift Planner around the way stations actually work
Abylon developed a web-based Shift Planner that brings schedules, absences and workload targets together in a single interface.
Role-based access allows different users to work with the information relevant to them, while an audit trail provides visibility into changes. English and Slovak were built into the first version, with the translation layer designed from the outset to support additional languages as the solution expands internationally. But the application itself was only one part of the story.
- One developer, Claude Code and a structured AI-assisted development process
The delivery team was deliberately lean: one project manager, one business analyst who also carried out manual testing, and one developer using Claude Code.
Rather than using AI simply to generate code faster, Abylon built a structured engineering process around it.
Coding standards, testing requirements and review responsibilities were defined at project level. Development followed a repeatable workflow:
research → approve the plan → write tests for key behaviour → implement → review → commit
This structure gave the developer a clear framework for working with Claude Code while keeping technical decisions and final approval firmly in human hands.
- Using AI not only to build, but also to challenge the solution
AI-assisted development extended beyond implementation.
Separate AI agents reviewed code quality, security and software architecture. A QA agent tested user journeys in a real browser and provided screenshot evidence when issues were identified, allowing fixes to be checked again against the actual application.
During the full-codebase review, an independent verifier also challenged serious findings before remediation work was prioritised. This added another layer of scrutiny instead of relying on a single AI assessment.
Automated pipelines enforced an 80% test coverage floor, while implementation plans and important development decisions were documented throughout the project.
The result was an AI-supported workflow designed not just for speed, but for consistency, traceability and maintainability.
Results
- A working MVP in 31 days
Just 31 days after the first commit, MOL Group had a working MVP. The first version already included:
- secure sign-in
- role-based access
- English and Slovak language support
- core workforce and shift planning functionality
- 590 automated tests
The MVP was not treated as the finish line. It became the foundation for the next stage of development.
- From 590 to 934 automated tests
Over the following three months, Abylon worked closely with MOL Group to incorporate additional requirements and refine the Shift Planner based on the evolving needs of the project.
Testing continued to expand alongside functionality. Internal agent-led security reviews and full-codebase reviews were completed, and by the end of development the application had reached 934 automated tests.
Automated deployment pipelines were also established to deliver the application into MOL Group’s Azure environment, supporting a more consistent path from development to deployment.
- What the project demonstrated about AI-assisted software development
The MOL Shift Planner project demonstrated what becomes possible when AI coding tools are combined with a disciplined engineering process.
A small delivery team was able to build and continue developing an enterprise workforce planning application without removing human responsibility from the process. AI accelerated implementation, testing and review, while defined standards, independent verification and human approval provided the structure around it.
The approach also produced value beyond the application itself.
Documentation created during development supported future maintenance and handover, while Abylon turned the development workflow into a hands-on AI-assisted software development workshop, enabling other developers to apply the same principles in their own projects.
For Abylon, the project was not simply about using Claude Code to develop faster. It demonstrated how AI-assisted development, automated testing and human engineering oversight can work together to make lean software teams significantly more capable.
Other Case Studies
Project Type
AI-Assisted Software Development
industry
Oil & Gas / Retail
Technologies USED
- Claude Code
- Microsoft Azure
- Web-based application
