Case studyConstruction & InstallationBouwworksLive since Dec 2025

Service tickets from 10 minutes to 10 seconds

How Dammers Bouw and AgentsLab added an AI coworker to the office team, integrated into Bouwworks, no parallel system.

0 sec

Per service ticket

0

Cases processed

Thousands

Service tickets per year

0%

Time reduction

98%

Reduction

From an average of 10 minutes per ticket to 10 seconds.

The office team has time again for the work that really matters: contact with clients and coordinating the service teams.

01, Situation

Repetitive entry work taking up the office team's day

Dammers Bouw is a mid-size Dutch construction company specializing in service and maintenance. Clients like housing corporations send in service tickets every day: complaints and reports from tenants that Dammers has to resolve on site.

Those tickets came in by email, often as a PDF attachment, in formats that differ per client. The office team carefully entered each ticket into Bouwworks: classify, read fields, key them in. Exactly the kind of work an AI coworker can support, so the team can spend more time on the substantive side of the job.

Thousands

Service tickets per year

10 min

Average time per ticket

~1.5 FTE

Capacity on admin

The motivation

A substantial part of the office team's time went to rule-bound entry work. Time we'd rather spend on the relationship with clients and the tenants they serve, and on coordinating the service teams in the field.

02, Solution

One AI coworker watching the inbox and filling Bouwworks

Continuous monitoring, classification, extraction and entry, only on exceptions does a person step in.

  1. 01

    Email classification

    The AI recognizes whether an incoming message contains a service ticket and filters out irrelevant mail from the stream.

  2. 02

    Extract from PDF and email

    Work performed, materials, time entries and customer details are pulled from the attachment and email body in structured form.

  3. 03

    Fill Bouwworks

    Structured data goes straight into the ERP, no middle layer, no parallel system.

  4. 04

    Send confirmation

    Once processing succeeds, a confirmation email goes out automatically. Admin stays current in real time.

  5. 05

    Human-in-the-loop on exceptions

    On fewer than 5% of cases a person fills in the gaps. The platform learns from every correction, patterns and preferences build up.

Bouwworks-native

The AI coworker runs straight in the existing ERP. No parallel system, no standalone OCR tool, no extra portal for staff.

Learning platform

Processing patterns and feedback are stored. The system gets stronger the longer it runs, fewer exceptions, less validation.

Sector-wide

The same architecture works on AFAS, Exact and Dynamics 365. Implementation typically takes six to eight weeks.

03, Results

Old situation vs. new situation, live production data

Measured across 462 processed cases between December 2025 and February 2026.

Handling time per case

~10 min

10 seconds

−98%

Human involvement

100%

4.36%

Drops over time

Exception rate

n/a

1.92%

Stable, low

Time for substantive work

Limited

~1 FTE extra

Client contact & billing

Note.The new numbers were hit in the first months of production. As the platform learns more patterns, exception rate and human involvement keep dropping. The freed capacity doesn't go away, it rolls into the next processes, starting with billing (service ticket to invoice).

04, Timeline

From scan to production in six to eight weeks

  1. Discovery

    Process review

    Mail flow mapped, formats inventoried, integration points with Bouwworks defined.

  2. Build

    Bouwworks integration

    Seamless integration into the existing ERP. No new infrastructure, no parallel system.

  3. Dec 2025

    Live in production

    First service tickets processed end-to-end. Inbox monitored, classification and extraction operational.

  4. Q1 2026

    462 cases processed

    Live data over the first months: 10 seconds per ticket, exception rate 1.92%, time freed for the substantive work.

  5. Now

    Billing in development

    Expansion: the service ticket plus the work performed flow on to the invoice, so admin runs from ticket through to billing in one stream.

What comes next

From service ticket to invoice. One continuous flow.

Processed service tickets, together with the work performed, will flow on to billing. So admin runs from ticket through to invoice in one stream.

BillingTicket to invoice

05, What this really means

An AI coworker on the team, not a replacement for it.

For Dammers Bouw the starting point was clear from day one: the office team stays, and stays in charge. The AI coworker takes over the rule-bound entry work, so staff have more time for the work only people do well: contact with clients, quality control and coordinating the service teams in the field.

The social aspectweighs heavily in this sector, both at Dammers and at their clients. That's why this partnership is set up as an extension of the team, not an alternative to it. The AI prepares, the human decides; exceptions are handled by staff and feed the system.

The same architecture works on AFAS, Exact and Dynamics 365, and the implementation typically takes six to eight weeks. The real work is not in the technology but in finding the right division between people and AI per organisation.

Service tickets or invoice flow?

Find out what an AI coworker can do for your team.

We start with a Quick Scan. No commitments, just a concrete plan on the table within two weeks.