Helm & Nagel GmbH has driven the AI revolution forward with Konfuzio in 2023: A dedicated marketplace, a lightning-fast tokenizer and smart upgrades such as Splitting AI and Bulk Edit - to name just a few. Supported by strong partnerships, we guarantee even more power for 2024.
Stay tuned: A number of innovations are already on the way!
You are reading an auto-translated version of the original German post.
Notable features and improvements in 2023
Marketplace
The marketplace fulfills the crucial purpose of introducing you to your AI journey quickly and easily. Thanks to it, it is within your reach to have a ready-made out-of-the-box model that you can use to process your documents without any training. It only takes a few steps before you can upload your own documents and extract the desired information from them. In addition to smoothly introducing you to the world of artificial intelligence, the marketplace is also designed to enable collaboration and sharing, fostering a vibrant community. As a user, you can share models that you have trained without making the data used public.
Increasing the performance of AI models
Significant performance improvement in training and deployment of all AI models released as of November 20. The improvement is particularly beneficial for large documents. More specifically, the Tokenizer is now up to 40 times faster.
Snapshot
The Snapshot functionavailable for local installations, allows you to save the status of any Konfuzio project - with all documents, models and configurations - at a specific point in time. Once saved, the Snapshot can be used to recreate the project in its saved state across server installations.
Splitting AI
The Splitting AI is an essential AI service that allows you to train a model to automatically split documents into a desired set of smaller files. For example, you can use Splitting AI to automatically split merged documents that contain several different files - or separate a specific page from a large document to reduce the number of pages that need to be processed for AI extraction. This service therefore saves you processing time, manual work and operating costs.
Bulk Edit
The function Bulk Edit allows you to manage several documents at once. You are able to change the data set (training, test, etc.), category and assignee of all desired documents with just a few clicks and thus optimize your project operations.
Training and evaluation protocols
Our models now have training and evaluation logs that help track the processes initiated when training a model on the training dataset and then evaluating its performance on the test dataset. Both logs are designed to enable the technical user to debug any problems.
Tokenizer
With the Tokenizer function allows you to add and manage tokenizers per label for extraction AIs. Based on the annotations for a specific label, Konfuzio helps you to find the appropriate regular expression that serves as a tokenizer. Adding such tokenizers - tailored to the data of a specific label - improves the quality of the extracted data.
API in the 3rd version
We have released a new version of our API: Version 3. The new version has clearly dedicated endpoints as well as new, more comprehensive functions such as updating, deleting and retrieving details of an annotation.
Closed partnerships in the annual review
Konfuzio is very proud of its growing presence in various industries, which is made possible by a robust network of partners. These partnerships not only provide direct customer support, but also actively resell Konfuzio to their customers. In addition, our key accounts have direct access to our in-house expertise, increasing the overall value we provide.
Find out more about the partners we have already introduced in blog posts:
Hoverture
Hoverture specializes in highly regulated industries such as finance and healthcare. The partnership will shape the digital transformation landscape in these industries by leveraging Konfuzio's technology to enable automated data mining, improve accuracy and accelerate decision-making processes. Hoverture will implement Konfuzio in its Google Cloud Platform environments, ensuring seamless integration and efficiency.
→ Read our blog post about the partnership for more details
Symplasson
Symplassona renowned IT service provider, uses Konfuzio's advanced AI technologies for automated and precise solutions in complex business processes and supports companies in their digital transformation. The collaboration, which also includes leading document storage and archiving provider M-Files, offers customized AI solutions that improve operational efficiency, reduce costs and increase customer satisfaction.
→ Read our blog post about the partnership for more details
Origoss Solutions
Origoss Solutions specializes in professional cloud services with a focus on the cloud native ecosystem. The partnership, based on advanced AI and cloud-native technologies, will provide unique solutions for companies looking to modernize, optimize and automate their operational processes. The collaboration will improve scalability and overall efficiency for customers on their digital transformation journey.
→ Read our blog post about the partnership for more details
People for AI
The partnership with People for AIa company specializing in annotation services, aims to accelerate the implementation of AI models for Konfuzio's clients by providing access to People for AI's expertise in annotation. People for AI is known for handling complex annotation projects in industries such as document processing, food classification, autonomous vehicles and aerial and satellite imagery.
→ Read our blog post about the partnership for more details
fino
The aim of the strategic partnership with fino (the creator of fintract) - a leading FinTech company - is to set new standards in intelligent automation in the financial sector. Standards to improve the efficiency and accuracy of financial work processes through automated digital solutions.
→ Read our blog post about the partnership for more details
Wolters Kluwer
The partnership with Wolters Kluwera global provider of professional information services, will improve the efficiency of administrative processes in public administration by introducing digitized workflows.
→ Read our blog post about the partnership for more details
apinity
Konfuzio and apinity - an API management platform - offer automated solutions developed exclusively for the insurance industry on the publicly available apinity marketplace. With these innovative solutions, companies can make their processes more efficient, reduce costs and offer their customers a better service.
→ Read our blog post about the partnership for more details
Notice: There are even more partnerships that were formed in 2023. You will soon find the blog posts on our website.
It's worth checking back regularly!
Are you interested in a partnership? You can find information about our partner program here.
Updates in 2024 - Stay up to date!
Following are our notable mentions that will enable you to achieve even more with Konfuzio in the year ahead - and with less effort:
- Improving the performance of categorization AIs.
- We are working on setting up the annotation of non-textual information. Currently, annotations always consist of text. When an annotation box is manually created, the Software will extract only the text detected within the bound of this box. Our upcoming change will create the ability to annotate any region within the document. This is a step towards the ability to extract and process non-textual information such as diagrams, images, etc.
- For our on-premise customers, it will be soon possible to use the Konfuzio server with an Oracle database instance instead of PostgreSQL. In combination with the Single Sign On this Oracle DB compatibility enables a smooth integration of Konfuzio into any corporate IT environment.
- Other functions that are based on LLMs to improve the handling of variance between documents and to interact with documents in a more flexible and sophisticated way.
- Introduction of experiment tracking for training models with a friendly user interface. Tracking provides comprehensive monitoring and analysis of your models' performance during training. This information can enable informed decisions to improve the training process.
Explore our section In Progress in the changelog for detailed insights into our ongoing developments.
Questions, comments or criticism?
Write us a message: