.svg)
A small but rapidly expanding health-insurance provider faced a bottleneck in one of its most critical — yet most manual — workflows: securing medical-record extracts for new claims. Each week the firm issues hundreds of record requests, and in a typical month those requests reach more than 150 different hospitals and clinics across Germany, Austria and Switzerland. When the documents arrive, claims agents still have to review, classify and file them before any reimbursement can begin. Four full-time employees were spending almost all their time chasing missing responses, sending follow-ups and re-typing data into spreadsheets — until they partnered with NOA.
The process posed a dual challenge rooted in its complexity and inconsistency. Each medical provider accepted record requests in its own preferred format: some clinics insisted on printed letters, others required encrypted emails with spreadsheet attachments, and a few offered their own web portals – often unreliable and difficult to navigate. Remembering “who wants what, and in which format” made standardization nearly impossible and turned a simple task into a logistical headache.
On top of that, the incoming documents were even more chaotic: scanned discharge letters, handwritten notes, ZIP archives filled with Excel files, poorly photographed PDFs — you name it. Extracting claim-critical information such as patient IDs, treatment dates, or diagnostic codes was slow, manual, and error-prone. All tracking was done using versioned Excel workbooks, which introduced frequent mistakes and made the process fragile; a single typo or misplaced row could delay an entire reimbursement.
By harnessing the power of advanced low-code tools and cutting-edge OCR-technology, the implemented solution eliminated the repetitive and cumbersome aspects of the process, empowering employees to focus on more meaningful, value-driven tasks. The core tools and steps that made this transformation possible are outlined below:
<aside> <img src="/icons/database_green.svg" alt="/icons/database_green.svg" width="40px" />
The first step in optimizing the process was transitioning data management from Excel to Airtable. Airtable’s structured, database-driven approach offered significant improvements, combining the robustness of a proper database with the ease of use of a spreadsheet. Its intuitive graphical user interface enabled non-technical employees to manage and oversee the process independently.
As tier Airtable Enterpise was chosen. On this tier, Airtable offers European data residency (primary servers in Frankfurt) and a signed DPA, so special-category health data stays inside the EU.
</aside>
<aside> <img src="/icons/pull-request_green.svg" alt="/icons/pull-request_green.svg" width="40px" />
Make.com served as the core of the solution, elevating the entire process to a new level. The complexity of the original workflow wasn’t rooted so much in cognitively demanding tasks requiring advanced AI, but rather in the sheer variety of process paths and formats involved. This is where Make.com excelled. Unlike any other tool, it efficiently managed the intricate web of process complexity through its deterministic automations.
With Make, claim requests were seamlessly routed to the correct medical provider to obtain medical records, tailored to their unique tools and formats. At the same time, Make handled the diverse and often inconsistent responses of the different health-care providers by pre-processing the incoming data - whether it was zipped Excel sheets, photographed pages, or scanned PDF files - and channeling it into the appropriate format. This pre-processed data was then sent to Airparser (discussed below) for OCR-based data extraction, streamlining the entire workflow.
Just like with Airtable, the Enterprise plan was selected for Make. On this tier, all automations run within a dedicated EU organisation, supported by ISO 27001-certified infrastructure and a GDPR-compliant Data Processing Agreement (DPA).
</aside>
<aside> <img src="/icons/search_green.svg" alt="/icons/search_green.svg" width="40px" />
With rapid advancements in AI research, OCR technology has significantly evolved in recent years, becoming a widely accessible commodity. Numerous tools now offer exceptional data retrieval capabilities, but our client needed a solution that met three key criteria: ease of setup, user-friendliness for non-technical employees, and cost efficiency.
Together with NOA, Mindee was selected as the ideal tool to meet these requirements. Its intuitive interface allows non-technical users to easily make adjustments without needing engineering support. Additionally, Mindee delivers outstanding performance, achieving a 96% accuracy rate in processing the provided documents. This high level of precision, combined with its simplicity and affordability, makes it the perfect choice for handling the OCR needs of this workflow.
For this project the Enterprise plan with EU hosting was chosen. On this tier, every document is processed exclusively on European servers, covered by a signed GDPR-compliant Data Processing Agreement (DPA) and backed by ISO 27001-certified infrastructure. Input files are encrypted in transit and at rest, log retention can be limited to 30 days, and data is purged automatically after parsing — ensuring full alignment with the strict requirements for special-category health data in Germany.
</aside>
<aside> <img src="/icons/invitation_green.svg" alt="/icons/invitation_green.svg" width="40px" />
To streamline both the sending and processing of physical mail, two specialised providers were integrated into the solution. LetterXpress manages outbound letters, while Pfennigparade scans and uploads incoming post. Both services provide ready-to-use APIs that connect directly with Make.com, enabling a fully automated end-to-end process.
Both vendors operate in full compliance with GDPR: LetterXpress runs on German servers, is ISO 27001-certified, and offers a formal Data Processing Agreement (DPA). Pfennigparade scans all documents in secure facilities in Munich, provides standard DPAs, and ensures that paper originals are destroyed in accordance with the Bundesdatenschutzgesetz (BDSG). These safeguards ensure that even sensitive medical correspondence is handled securely and in line with strict health-data regulations.
</aside>
The automation solution delivered by NOA provides our client with several key benefits:
REDUCTION IN WORKFORCE REQUIREMENTS
Previously, managing the original, highly manual process required the full-time efforts of four employees. Thanks to the sophisticated end-to-end automation solution, the workload has been reduced to the equivalent of just half an employee. This remaining role primarily involves monitoring the automated system, with intervention only needed in rare edge cases.
IMPROVED DATA QUALITY
As mentioned above, Airparser achieved an impressive 96% accuracy in retrieving data from processed documents. Combined with faster processing times and more structured data management, this significantly reduced the occurrence of idle or incomplete data points, ensuring a smoother and more efficient workflow!
.svg)
FASTER “TIME-TO-PAYOUT”
The average time-to-payout — the duration from registering a new insurance claim to completing its full processing — was reduced from 21 days to just 9.5 days. This acceleration led to a noticeable improvement in customer satisfaction and operational efficiency, freeing up valuable resources that could be reinvested into business growth and daily operations.
.svg)
ERROR MINIMIZATION AND ENHANCED EFFICIENCY
Automated document processing significantly reduced errors related to handling insurance claims and retrieving medical record extracts. Thanks to structured workflows, clear automation logic, and the elimination of most repetitive tasks, employees were able to manage requests and responses far more efficiently and accurately.
The decision to partner with NOA and leverage low-code tools for automating the process delivered outstanding financial results for the health-insurance provider. Here’s a breakdown of the costs, savings, and ROI: