Case study
Enterprise Management System
The central business platform of a group of companies: customers, tickets, offers, orders, invoicing, projects and time tracking, with AI built into everyday work.
- Role
- Software Engineer, now AI Software Engineer
- Years
- 2024 to today
- Source
- Private, built at work
01
One platform for the whole business
The platform runs the daily business of several group companies, each on its own instance: customers, tickets, offers, orders, invoicing, projects, time tracking and reporting. It is a long-lived Yii application, so the work is as much about modernising what exists as about building what is new.
I have developed and maintained it since 2024, first part-time and now full-time with a wider scope: bringing AI into the platform and building new modules. The code is private, so this page describes the work without screenshots.
Stack
- Yii Framework
- PHP
- MySQL
- JavaScript
- DATEV API
- Bootstrap
- LLM APIs
02
What I built
DATEV and e-invoicing
REST integrations including DATEV-related workflows, and ZUGFeRD XML generation for exchanging finance documents.
AI ticket assignment
Suggestions for who should take a ticket, with access control enforced on the server.
AI-assisted search
A search across customer history, covering tickets and order items.
HR and time tracking
An in-house module that replaced an external time-tracking tool: clock-in and clock-out with breaks, time accounts with target hours, public holidays and statutory break rules, a live attendance overview, logged corrections, and vacation and absence management.
Reporting and performance
Reporting, export and dashboard features, and faster pages through query optimisation and the removal of N+1 queries.
Modernisation
Legacy areas refactored into object-oriented code with automated tests and continuous integration, a telephony integration, and a move to a Bootstrap 5 admin template.
03
How the platform is organised
1
People
- Staff of several group companies
- Role-based access control
2
Modules
- Tickets, offers, orders, invoicing
- Projects, time tracking, HR
3
AI features
- Ticket assignment suggestions
- Search across customer history
4
Integrations
- DATEV and ZUGFeRD
- Telephony and REST APIs
Each group company runs its own instance of the same application.
04
The hard parts
01
AI that respects permissions
The AI features are access-controlled on the server, so a suggestion or a search result never shows more than the person is allowed to see.
02
Time tracking that follows the rules
Time accounts handle target hours, public holidays and statutory break rules, and every correction is logged.
03
Changing a system people rely on every day
New work ships next to legacy code, so changes are protected by code reviews, automated tests and concurrency-safe database transactions.
Next case study
Streaming Ad Platform
Want to talk about a project like this?
Get in touch