Sometimes great ideas come straight from everyday life. That was exactly the case for us: in 2017, we needed an internal system to manage holiday requests, time tracking, and team coordination all in one place and with local data storage — so we quickly decided to develop our own tool: TeamTracker. Back then, our main focus was simply on making it practical for our own use. The focus to extend the program to the needs of other companies came later.

Key features at a glance

• Calendar & holiday management: holiday requests are entered directly into the calendar and approved by the appropriate decision-makers
• Time tracking: In addition to manually recording working hours, we have now also implemented a check-in/check-out feature
• Team & project organization: A central repository to manage and maintain all project, team and customer data
• Integrated billing: Fully automated generation of time sheets and invoices
• Asset management: Create and manage company assets and loans (e.g., hardware or licences)
• Completely secure data storage and full control over data protection thanks to on-premise operation on the company’s own server

Our first client, a leading ophthalmology practice in Hamburg, was especially interested in the time tracking feature. In particular, they wanted a check-in/check-out function, which we custom-built for them. We also made the interface design fully customiseable to match the practice’s colours and logo — so the team can find their way around quickly and the tool visually fits the company.

Working with our first external client showed us how important it is to tailor TeamTracker to individual customer requirements. Contact us if you have individual needs for your time tracking system. We’ll continue developing the tool based on user feedback, placing special emphasis on an improved mobile version.

The winter season is a special time at LangTec, and this year’s Holiday party was no exception. Held on December 5th, the day before Nikolaus, we gathered to celebrate the year’s achievements and enjoy each other’s company in the heart of Hamburg.  Our celebration began at Miniatur Wunderland, an awe-inspiring attraction located in Hamburg’s historic Speicherstadt. Known as the largest model railway system in the world, it’s no surprise that this Guinness World Record holder is one of Germany’s most beloved sights.  During a behind-the-scenes tour, we were amazed by the logistics and technology that bring this miniature world to life. From tiny planes taking off to bustling cityscapes, every detail was a testament to the incredible creativity and dedication behind the exhibit.

After immersing ourselves in the wonders of Miniatur Wunderland, we took a leisurely walk to Die Perle Restaurant, a hidden gem in the Kontorhausviertel. True to its name, Die Perle shines as a “pearl” of Hamburg’s culinary scene. Known for its modern take on global dishes, the menu features a tempting array of seafood and truffle specialties that delighted our palates.  We dined like royalty, and as always, left full to the brim with good food, good conversation and a rousing round of Semantle.

We wrapped up the night feeling grateful for each other and another fantastic year at Langtec.
Here’s to a festive season filled with joy, and to an exciting year ahead!

While we are already supporting a large European insurance company with AI consulting, we now proudly announce the start of yet another AI project with a second global insurance giant.

In this new project, we are going to assist our client with their AI-based text management and content creation. The focus if this engagement is to utilise the possibilities of AI to facilitate the creation, editing and quality assurance of textual content more efficient. During text generation language models shall take into account the wealth of company-specific information and requirements regarding text and language style.

With more than a decade of experience in language technology and text automation LangTec advises in the design of such tools and their implementation, hand in hand with our client’s in-house AI team.

We look forward to an intensive and stimulating collaboration over the next six months!

As summer slowly faded into autumn, the LangTec team set out on a team-building adventure that embraced both Hamburg’s rich history and its modern transformations. The weather held out just long enough for us to complete our climb and experience one of Hamburg’s most unique landmarks, the St. Pauli Bunker.  The bunker, which has been recently transformed into the “Green Bunker,” offers a rare combination of history, architecture, and nature. Open to the public since July, its rooftop garden—the highest in Hamburg—features 23,000 trees, shrubs, and plants spread over five pyramid-like floors. Despite the gray skies, the panoramic views of the Elbphilharmonie, Michel, and the harbour were as stunning as promised.

After our hike, we settled into the bunker’s café Constant Grind for a cozy break, where we played the board game “Concept.”  The game, where you must guess words through the association of icons, seemed almost designed for a group of programmers and linguists like ours. We deconstructed meanings, toyed with symbols, and uncovered connections.

Finally, we ended the evening with a 3-course mystery dinner at La Sala, the restaurant inside the bunker. The meal was a surprise, and each course offered something new and delicious. It was the perfect way to wrap up an eventful, drizzly, yet memorable day.

It was with these words that our client entitled the e-mail in which they informed us about having won this contract. After months of intense alignments we are delighted to have been awarded the contract for the development of two new document parsers in the shipping and sea freight sector. Specifically, we will provide a customised instance of DocumentReader and operate it for our client. DocumentReader is LangTec’s custom solution for the high-precision extraction of business-relevant information from documents in any type of format. As a cloud-based service, DocumentReader integrates seamlessly into any digital process.

Our customer is a world-renowned, publicly listed company and one of the top logistics companies worldwide. We are super-happy about this cooperation and also proud of the fact that the DocumentReader’s amazing extraction precision has convinced our customer to partner with LangTec for this exciting project. Watch this space for more updates.

LangTec has taken over a consultancy for the central IT services management of a leading multinational insurance group. The purpose is to introduce automation and artificial intelligence (AI) for improved efficiency in ITIL-compliant business processes. The scope comprises both the transfer of knowledge about AI generally and application to specific recurring use cases as well as medium-term planning of further development.

The mandate specifically targets process optimization via AI in change management, incident response, licence administration and vulnerability management. All these business cases are currently implemented in a standard ERP platform with extensive customizations and are subject to considerable regulatory constraints. Although central configuration management database ultimately contains all relevant information about persons, groups, devices, licences, tasks etc., these are often not available in the necessary context of the daily tasks.

LangTec will tackle challenges such as: automatic evaluation of incoming Change Items (Is the risk evaluation meaningful and accurate? Are all affected work groups linked?) or Incident Tickets (Are all relevant descriptions present?), data visualisation (dashboards displaying the status of incidents or aupporting auditing by management) or aggregated evaluation of the current status (Can an imminent major outage be recognized early?).

We look forward to the next months of intensive and engaging cooperation!

LangTec is excited to announce its latest collaboration in data engineering with a globally leading chemical and consumer goods company headquartered in Düsseldorf, Germany. The focus of this project is on enhancing our client’s data infrastructure. As part of the project, LangTec will play an integral role in the development of a comprehensive data catalogue and data mesh, ensuring the reliability and governance of data sources across various departments within the organisation.

The project involves the provision of expert guidance on data architecture and connections, implementing and constructing efficient data pipelines, and performing essential data transformations to meet the diverse business needs. This comprehensive approach enables our client to harness the full potential of their data assets, driving informed decision-making and operational efficiency.

This client project underscores LangTec’s commitment to delivering cutting-edge data solutions, fostering innovation, and making a contribution to the empowerment of large-scale organisations to optimise their data assets and processes.

This article examines the performance of large language models (LLMs) like ChatGPT4-Turbo and ChatGPT4-Omni for information extraction tasks, comparing them with LangTec’s specialized E-MailParser. Our analysis reveals significant limitations of LLMs in this domain.

Benchmarking Accuracy Scores

To benchmark the accuracy of ChatGPT4-Turbo, ChatGPT4-Omni, and LangTec’s E-MailParser, we conducted a comprehensive evaluation using 20 documents across four extraction tasks:

• Q88: Vetting Questionnaires for Tanker Information
• Timesheet: Tanker Loading/Unloading Statements of Fact
• Ship: Requests for Commercial Cargo Shipping
• Cargo: Commercial Cargo Vessel Position Lists

From each of these documents we extracted about 20 target data points. For the evaluation,These documents had predefined ground truth labels, indicating the expected target values for each field. By comparing the extracted values to these ground truth labels, we were able to calculate accuracy scores for each model.

Benchmark Results: LLMs vs. LangTec’s E-MailParser

Both ChatGPT4-Turbo and ChatGPT4-Omni show some level of accuracy in information extraction tasks, achieving overall scores of 56 % and 49 % respectively. Notably, the newer model ChatGPT4-Omni performs worse on this task for most document types than its predecessor ChatGPT4-Turbo. Another important observation was that model performance is impaired by inconsistency. For the same input text and extraction task, these models provide different answers each time they are queried, even when prompted for the same question. This non-deterministic behavior renders them unreliable for scenarios where consistent and accurate information retrieval is essential.

In contrast, specialized parsers such as LangTec’s E-MailParser exhibit significantly higher accuracy, are fully deterministic in their behaviour and consistently achieve 98 % extraction accuracy across various document formats. This reliability makes a deterministic document-understanding solution like E-MailParser a more dependable solution for information extraction tasks, particularly when dealing with diverse e-mail content in business-critical applications.

Conclusion

While LLMs like ChatGPT are excellent for generating content, they have notable limitations, particularly in scenarios requiring deterministic output such as information extraction tasks. For such applications, document-understanding solutions like LangTec’s E-MailParser offer a more reliable and accurate solution.