Case Studies

We deliver value to make a difference

Learn how we collaborate with our clients and create innovative end-to-end solutions.

Our three-step approach

Step 1

Your domain & pain

We will always understand your side of the story first. This is how we ensure our solution bring true value to the table.

Step 2

Tailored solution

Our architects create a solution that fits your users needs and constraints. We identify the ideal team constellation and source it.

Step 3

Delivery & iteration

We make sure to fully deliver your product and iterate as needed and provide support, maintenance and training.

All of our services – be it AI powered Software, conventional tailored Software, or AI Consulting and workshops – go by this approach. We make sure to find exceptional engineers and bridge the communication and cultural gap in your favor.

AI-Powered Proposal Processing with Antrag AI

Problem

DOMCURA faced a backlog of over 10,000 manual PDF insurance applications, which significantly delayed policy issuance. The manual processing of these proposals consumed resources and slowed down operations, affecting overall efficiency. This bottleneck hindered timely service delivery and created challenges in scaling operations effectively.

Solution

HEON developed Antrag AI, an AI-driven solution that automated proposal processing using LLMs and OCR. The system classified applications, extracted policy-relevant data, populated forms, and initiated policies, with uncertain cases flagged for human review. This approach ensured both high accuracy and compliance through integrated audit logs and transparent workflows.

Outcomes

The backlog was reduced by 90%, from 10,000 to 1,000 proposals, enabling quicker policy issuance. Average handling time dropped from 30 minutes to under 5 minutes, significantly improving operational efficiency. The solution enabled a measurable increase in policy creation while maintaining transparency and traceability in decision-making.

Having a similar problem?

AI-Powered Email Processing with Bestand AI

Problem

DOMCURA struggled with processing 1,500–2,000 emails daily, relying on outdated software that correctly classified only ~10% of emails. This inefficiency not only burdened the team with excessive manual sorting but also lacked the scalability required to handle high volumes. The process cost the company €60,000 per quarter and delayed customer responses, impacting service quality.

Solution

HEON introduced Bestand AI, a tailored AI solution that automated email classification and streamlined the handling of frequent requests like address changes and cancellations. The system, integrated with advanced AI tools, achieved classification accuracy above 50%, far exceeding the performance of previous software. This capability drastically reduced manual effort while ensuring accurate and timely processing of customer requests.

Outcomes

Email classification accuracy soared to 50%+, replacing the inefficient legacy software. DOMCURA realized substantial savings on the €60,000 quarterly cost of ineffective solutions. The faster processing and routing of emails improved customer response times and operational efficiency, enhancing service delivery.

Having a similar problem?

AI Vision Unlocks Form Processing at Scale

Problem

MLP faced challenges during a global product configuration update, as customers submitted thousands of scanned PIN/TAN forms daily, many of which were of poor visual quality. Legacy OCR tools failed to extract critical data, particularly checkbox selections, which disrupted data processing workflows. This created an urgent need for a scalable, accurate solution to ensure operational continuity and reduce manual workloads.

Solution

HEON designed a hybrid AI system combining OCR, image enhancement, and LLMs to automate form processing with high precision. The system transformed low-quality scans by dynamically enhancing checkbox areas and merging textual context to interpret checkbox selections with over 90% accuracy. This structured output supported downstream automation, ensuring robust, high-throughput document processing workflows.

Outcomes

MLP achieved >90% accuracy across thousands of real-world PIN/TAN forms, enabling seamless data extraction even from poor-quality scans. The batch-processing pipeline reduced costs to <€0.01 per document, delivering a massive ROI compared to manual or legacy tools. High-speed processing, daily quality tracking, and explainable decisions enhanced reliability and compliance with regulatory requirements.

Having a similar problem?

AI-Powered Accounting Transformation

Problem

Maute & Schrader faced inefficiencies in their accounting workflows caused by the limitations of classic OCR tools, which failed to interpret the relationship between invoice details and their accounting implications. As a result, extensive manual intervention was required, leading to slower processing, higher error rates, and reduced scalability. Despite a successful proof of concept (PoC), the firm lacked the resources to push the solution to production readiness, further delaying progress.

Solution

HEON developed a context-aware Accounting AI tailored to Maute & Schrader’s unique needs. The solution combined OCR and Vision AI for invoice detail extraction and analysis, while leveraging client-specific customizations to enhance categorization accuracy. Automated classification, rule-based consistency checks, and seamless integration with accounting tools like DATEV streamlined invoice processing and ensured compliance with industry standards.

Outcomes

The proof of concept achieved 88% classification accuracy, demonstrating the effectiveness of AI in tackling nuanced accounting tasks. Manual intervention and rework were significantly reduced, improving scalability and efficiency for handling high invoice volumes. Enhanced transparency, consistency, and audit readiness positioned Maute & Schrader to future-proof their operations with scalable automation.

Having a similar problem?

AI-Powered Transaction Processing

Problem

FERI faced significant inefficiencies in managing depot transactions, manually processing thousands of PDF statements weekly. This resource-heavy approach hindered scalability, increased costs, and posed risks of data errors. Ensuring compliant, traceable associations between extracted data and internal records further complicated the process, slowing down operations.

Solution

HEON developed an AI-powered system that automated PDF classification and data extraction with over 95% accuracy. The AI efficiently mapped client and depot numbers, transaction amounts, and other key data to FERI’s internal records with full transparency and traceability. This scalable solution was seamlessly integrated into production, empowering FERI to reduce manual workloads and independently manage future enhancements.

Outcomes

FERI achieved over 95% accuracy in processing depot transaction PDFs, eliminating the need for most manual interventions. Scalable workflows significantly reduced operational costs, saving time and resources for high-priority tasks. Enhanced traceability ensured compliance with financial industry standards, supporting FERI's sustainable growth and operational efficiency.

Having a similar problem?

Helping push the renewable energy revolution

Problem

A joint venture between Shell and NWG Group intends to secure 10% of DACH's EV charging station and solar panel operation market share. They need fully automated business processes and customer portals to manage their rapid growth.

Solution

In close collaboration, we designed and built a bespoke, cloud-first product suite to meet their needs. It includes a product for storing millions of real-time metering sensor datapoints, one for real-time synchronization between various IT systems, and an integrated customer and partner portal.

Outcomes

The company can now confidently engage with major players in the German market, leveraging automation for data and payment processing. The product suite has enabled the company's growth, effectively managing large volumes of data and traffic.

Having a similar problem?

Revolutionizing professional sports

Problem

A markerless motion tracking system for live professional sports events was developed, necessitating updated on-site application management, tracking processes, and a new analytics portal due to the system's success. The increase in data and tracked events presented significant challenges to maintaining high data quality.

Solution

We established a data warehouse to consolidate tracking data and sessions, generating daily quality reports and alerts. We formulated a containerization and update strategy specific to modern sports facilities, enhanced existing high-performance compute pipelines to accommodate new product requirements, and helped design a versatile analytics portal to process various sports and movement data.

Outcomes

Originally designed for sports labs and professional baseball, the product is now being successfully deployed in numerous contexts, including professional soccer, basketball, fencing, and Olympic events.

Having a similar problem?

Creating a legal and compliance data hub

Problem

Major corporations needed an accessible, user-friendly method for employees to engage with compliance information, usually compiled by a law firm but often neglected due to its storage in hard-to-reach areas. Employees not having access to such information was a huge risk for triggering non-compliance issues.

Solution

We developed a CMS-based progressive web application that is easy to use and capable of offline operation. It features advanced search capabilities, robust security, and efficient user management, enabling quick access to relevant information.

Outcomes

The web app has transformed the compliance landscape within these organizations. The user-friendly design and advanced search functionality have led to increased usage and heightened awareness of compliance matters. The decrease in non-compliance issues and the continuous, offline availability of vital information have fostered a proactive compliance culture across the enterprises.

Having a similar problem?

Helping doctors find better treatment for rare liver diseases

Problem

Despite the range of rare liver diseases, few treatments exist. To address this, the EU has sponsored years of decentralized data collection across European medical universities. To utilize machine learning, this data needed merging into a single schema.

Solution

We partnered with internal engineers and medical staff to define a common schema iteratively. Then, we automated the target schema creation from the agreed-upon human-readable specification and executed the data migration.

Outcomes

This successful data consolidation into a unified schema has advanced analysis capabilities, paving the way for machine learning models that support more accurate diagnoses and treatments. It has spurred medical research and offers promise for new targeted therapies for diverse rare liver diseases.

Having a similar problem?

Building the next generation social volunteering platform

Problem

A non-tech startup required a complete overhaul of their corporate volunteering platform to meet enterprise needs, including a novel chat-based system for linking volunteers and nonprofits.

Solution

Following several event storming sessions, our team created a cost-effective, event-driven, data-centric solution, making use of third-party services. After confirming the user flows via usability tests, we assembled a team of expert engineers for implementation.

Outcomes

We flawlessly launched the platform to multiple enterprise clients, onboarding thousands of users. We then transitioned to economical maintenance mode and prepared for handover to a future in-house CTO.

Having a similar problem?

Overcoming the financial hurdle for forwarders in the logistics sector

Problem

In logistics it is common practice to pay the forwarder only after (t30+) receipt of the goods, while most of the costs for transportation are incurred directly. Especially small businesses suffer from this discrepancy. Often they have the capacity, but there is a lack of liquidity.

Solution

With the Kentra start-up we aimed to close the gap by connecting forwarders and funders and enabling respective lending in a secure and well-rounded way. We developed an API in which the forwarder can apply for loan via his order invoice. The calculation is documented and well prepared for the funder, which enabled next-day transactions.

Outcomes

Due to framework agreements concluded in advance, forwarders got payed loans from their funders within 24h. The fast procedure allowed the them to accept proposal and helped them to consolidate their position in the business.

Having a similar problem?

Already intrigued?

Let's see how we can be of help !

"The project has completely transformed our claims workflow. We’ve seen a 50% reduction in manual processing and a marked improvement in customer satisfaction."

Ralf Maute , Founder - Maute & Schrader

"Thanks to the new system, our underwriters now focus on high-value assessments instead of chasing documents. The ROI speaks for itself."

Lars Malinowsky , Head of IT - Domcura AG