OneStop Pro
90 % less manual document work: how OneStop Pro leveraged Klaaro in a pilot project
Up to 60,000 service and maintenance invoices a year, with varying layouts and inconsistent line-item names: for one use case, OneStop Pro was looking for a way to automatically process the relevant data from these highly heterogeneous documents. In a multi-week pilot phase, the team tested Klaaro and, based on the results, estimated a potential to save 90 percent of time spent on document workflows in production use.
Tens of thousands of invoices a year – no human can keep up with that
OneStop Pro (“OSP” below) is a software provider for digital equipment management and tracking of machines, equipment, and operating resources in the construction industry. The company helps bring all information about an equipment fleet together in one place.
While digitizing equipment management and tracking, OSP recently faced large volumes of service and maintenance invoices from various service providers. Because these vary widely from provider to provider in layout, terminology, and line items, conventional automation tools would fail on the sheer variety of documents. Plain OCR (Optical Character Recognition) wasn’t the right fit either: it digitizes documents, but doesn’t process them automatically. For a specific use case, OSP was facing around 60,000 invoices a year. Processing them manually was quickly ruled out because of the time involved, the monotony in execution, and the error rate. What OSP needed was a solution that would take over the entire process of transforming inconsistent documents into structured data, from intake all the way to the target system.
From a meetup to a successful pilot phase
In March 2026, the AI meetup Niederbayern AI took place at the INN.KUBATOR in Passau. There, Julien Kollmann of synsugar presented the company’s new product, Klaaro, which was still in development at the time. Core features designed to ease manual document work through AI-powered data extraction already existed but were not yet in production use. Through this presentation, Sebastian Dillinger, Product Owner at OneStop Pro, first heard of Klaaro and subsequently got in touch with synsugar.
The original idea for integrating Klaaro into OSP’s processes concerned a different use case: handwritten documents filled in by construction workers on site should be digitized more easily, read automatically, and processed directly in the target system – something Klaaro could already do. After a product demo shortly afterwards, however, an even bigger use case came into focus.
It concerned around 60,000 service and maintenance invoices that would have had to be processed manually because of their complexity and heterogeneity from one service provider to the next. For example, the same line item is labeled differently by each provider, which would complicate any automation and result in manual effort for OSP.
To tackle this effort, the question quickly arose whether Klaaro could cover this use case as well. After the promising demo and an in-person visit to OSP, synsugar and OSP quickly agreed to test Klaaro on the new use case in a pilot phase over the course of several weeks.
The pilot phase officially started with a kick-off meeting at the startup center where synsugar is based. Implementing Klaaro and setting up access went quickly and smoothly. The early stage of the pilot project was mainly about consulting by synsugar and optimizing the data schemas, on the basis of which OSP was able to process more and more of the service and maintenance invoices in a fully automated manner.
Klaaro itself also evolved during the pilot phase. While parallel processing of a manageable number of documents already worked flawlessly at the start, the pilot project made it possible to scale this up to any number of documents. In one test, fewer than one percent of 2,000 documents processed in parallel returned an error, and in every case a second run of the extraction pipeline resolved it.
With Klaaro we extended our tech stack to unlock valuable AI features. Thanks to automatic document processing, we can reduce the time spent on handling documents manually by up to 90 percent.
The results: structured invoice data and significant savings potential
Unlike extraction with a single LLM (e.g. ChatGPT or Claude models), the data extraction in Klaaro remained reproducible across thousands of different documents and added no time overhead, thanks to parallel processing of documents.
Particularly because Klaaro automatically generates a suitable data schema for each individual document type, OSP was able to run its first extractions within a very short time. Since Klaaro also orchestrates non-deterministic LLMs in the background, it was still important for OSP to refine the data schemas itself, fine-tune them more closely to their data requirements, and supply relevant context in each case to make the data extraction as accurate and reliable as possible.
With additional consulting from synsugar to improve the data schemas, OSP was also able to quickly ensure remarkable data quality. The line items listed on the service and maintenance invoices ended up reliably in the schema’s output format in full detail, and could be passed straight into OSP’s own service via the REST API.
At the end of the pilot phase, the results were evaluated once more, and based on the extractions and data quality, the time savings for a production rollout of Klaaro were assessed. OSP estimated that eliminating manual document work in this use case would save them 90 percent of time.
