How a B2B car rental company serving industrial and premium clients transitioned from a standard software that couldn't reflect its reality to a learning Operating System – now proactively managing vehicles, orders, and prices instead of merely administering them.
Initial Situation
The company rents vehicles to industrial clients and in the premium segment – often project-based, with changing requirements for vehicle types, durations, and availabilities.
A business that thrives on fast, precise scheduling. Formally, there was a standard software in place.
However, the everyday reality looked different:
- Central information was stored in free text fields
- The team organized processes outside of the system – in spreadsheets, via phone, WhatsApp, or in the minds of individual employees
- The scheduling team received only superficial support from the system
- Historical data was unstructured and inconsistent
The software was not a control tool. It was a compromise. The team had learned to work around the system. It worked – but it consumed capacity, generated errors, and made growth a risk.
The standard software did not reflect reality – the team had to piece together their reality around it.
Three Stages to Operational Freedom
We structured the project along our 3-stage model. Each stage builds on the previous one. After 16 months, all three stages are operational.
| Stage 1 – Order | Stage 2 – Automation | Stage 3 – Freedom |
|---|---|---|
| Cleansing legacy data, clean modeling of vehicles/orders/drivers, building the core scheduling | Streamlined scheduling processes, digital handovers, status logic | AI availability forecasts, proactive deployment planning, dynamic pricing |
Stage 1 – Order
Before we could automate anything, we had to rebuild the system foundation. The standard software and its Excel companions were not suitable as a foundation. In Stage 1, we established:
- A clean modeling of core objects – vehicles, orders, procurement channels, customers, drivers
- Clear status and accountability logic – when is a vehicle reserved, when is it issued, when is it returned, who decides what
- A central scheduling system that reflects real workflows, not a generic average process
- A differentiated migration – we carefully transferred the relevant core information into the new structure
In doing so, we made a conscious decision: we did not adopt unusable data junk – inconsistent entries, orphaned free text fields, process logics that no one understood anymore. Better a clean core than a burdened migration.
The bravest step was not the new system. It was the willingness to truly let go of the old one.
The implementation was controlled: Workshops with the operational teams. Key users from scheduling took on multiplier roles. We gradually activated the modules and maintained a temporary parallel operation for the most critical processes. We did not abruptly introduce the new system – we established it cleanly.
Stage 2 – Automation
Based on the structural clarity we achieved, we purposefully added automations. The company saw the greatest effect in scheduling: Previously, each morning began with the same routine – opening standard software, checking multiple Excel spreadsheets, reviewing incoming emails & WhatsApp messages, manually cross-checking vehicle availabilities, assigning customer inquiries individually, and informing drivers by phone.
Duration: about four hours every day.
Today, scheduling opens a system that already knows the status. Vehicles, orders, and drivers are interconnected. Status changes automatically trigger follow-up processes. Driver assignments are managed digitally. Handoffs are documented by the team on-site at the vehicle. Tasks are created systematically – no one has to compile them first. From four hours, it became 40 minutes.
Not a theoretical value, but the actually measured difference in day-to-day business.
We intelligently replaced the legacy system
We no longer operate the old system. Instead, we differentiated: the relevant data has been migrated into the new structure, the complete database dump is archived, and a lean archiving solution makes historical processes accessible at any time. An AI-supported search function scans the archive in natural language. The old system is gone. The knowledge from it remains – without ongoing (licensing) costs.
Stage 3 – Freedom
With a stable data foundation and functioning automation, the system has reached a point where it can share responsibility for strategic questions. Stage 3 has changed how the company manages its fleet.
AI-Powered Availability Forecasts
The system now predicts which vehicles will be available in the coming weeks – and when. Project extensions, typical return times, seasonal patterns, and customer-specific usage behavior are taken into account. Scheduling not only sees the current situation but also how the fleet status will look in three, six, and twelve weeks. With new inquiries, the system knows whether and when the appropriate vehicle will be available – without anyone having to check in Excel.
Proactive Planning
When an inquiry comes in, the system automatically suggests the optimal vehicle-project allocation. Criteria include: availability, vehicle profile, location, customer preference, contribution margin. The scheduler reviews and confirms – without having to search themselves. Fleet utilization has measurably increased, even though scheduling time is decreasing.
Dynamic Pricing
The system continuously adjusts rental prices – based on demand, seasonality, vehicle availability, and customer segment. Management centrally defines rules and limits, and the system executes them. Margins per rental have increased without losing customers. For industrial clients with framework agreements, the agreed conditions remain intact; the dynamics take effect where they should.
Proactive Prioritization and Early Warning
The system alerts when critical patterns emerge – bottlenecks in certain vehicle classes, projects that historically overrun, customer inquiries that pose risks in combination with the current fleet status. Scheduling receives alerts before a potential shortage becomes an outage.
AI Agent for Offenses
With a fleet operating across Europe, mail from authorities arrives every week – violations, hearing forms, ownership inquiries from various countries, in different languages and formats. Previously, an employee would open each letter individually, translate it, look up the owner in the system, and manually fill out the response form. As fleet volume increased, this became a noticeable time sink – and simultaneously a risk of errors if deadlines were missed.
Today, an AI agent handles the entire process. Incoming letters are scanned – regardless of language, layout, or sender.
It identifies the time of the offense, location, license plate, and accusation, cross-references this in the system with rental contracts and driver assignments, and identifies who was driving at the time of the offense. The reply is filled out by the agent in the respective format of the municipality or police authority – paper form, online portal, or email response. It is returned via the channel it came through. An employee only needs to review what the agent could not clearly allocate. What once took hours of work per week has now become a few minutes of checks.
Measurable Results After All Three Stages
| Area | Before | After Stage 2 | After Stage 3 |
|---|---|---|---|
| Scheduling Time | 4 hours daily | Approximately 40 minutes | Primarily reviewing and confirming, no longer searching |
| Order Processing | Manual, fragmented | System-driven and seamless | Proposals are created automatically, scheduling decides |
| Offenses | Manually – hours per week | Structured driver assignments in the system | AI Agent takes over completely – employee only verifies |
| Error Rate | High due to communication breaks | Significantly reduced | Early warning of critical patterns before errors occur |
| Availability Forecast | None – only current stock | Current stock, well-maintained | Forecasts 3, 6, and 12 weeks ahead |
| Pricing | Uniform, rarely adjusted | Pricing rules in the system | Dynamic, per segment, continuous |
| Legacy Access | Ongoing licensing costs | AI-supported archive search | AI archive search as an integrated component |
| Control Mode | Reactive – after the fact | Proactive – with early signals | Proactive – the system anticipates |
Up to 80% time savings in scheduling – while simultaneously increasing fleet utilization and rising margins per rental.
What Has Changed for All Stakeholders
The system has changed the way the company works. Not only operationally – but strategically as well. Scheduling no longer searches for answers – it makes decisions. Schedulers review system proposals, confirm or adjust them, and actively manage the business rather than just administering it. Fleet managers see where utilization stands, where vehicles are underperforming, and where investments in new vehicle classes are worthwhile. Sales provides customers with reliable statements on availability for inquiries – in real-time, even for points several weeks into the future. Management plans fleet growth based on real forecasts, not gut feelings.
Previously, the day consisted of searching, cross-checking, and calling. Today, it consists of making decisions.
And perhaps most importantly
The company has decoupled its growth from its coordination efforts. More vehicles, more customers, and more parallel projects do not make scheduling more complex – they make the system more precise.
What This Project Shows
This project showcases a pattern that occurs in many operationally complex companies: Software exists that is formally supposed to cover the tasks. However, it does not suit practical needs – leading to workarounds in Excel, email, and phone calls. The team becomes accustomed to this, and eventually, they no longer notice how much capacity is flowing into bridging the gaps. The first step was not to find even better standard software. It was the decision to build a system that fits real working methods – and the courage to truly let go of the old system. Only this order has enabled automation. Only this automation has enabled freedom.
Order enables automation. Automation enables freedom.
Anyone working with software that does not reflect their reality should ask themselves a question: Does my system reflect my processes – or do I shape my processes around my system? If the answer is "I build around it," then the leverage is not in another standard tool. It lies in a system that lives with the business – and in what becomes possible when it is in place.