KI-Lösungen - t2b.ch

AI-supported Solutions

Your Data. Your AI.

Secure. Data stays in your data center.
Efficient. thalabus has low infrastructure requirements.
Fast. AI solutions are created in a few weeks on thalabus.
Correct. Your answers are based on your data and the source can be displayed.

Chatbots for Customers and Employees

Multilanguage chatbots are the "classic" realm for Large Language Models LLMs. Whereas it is easy to create a Chatbot for simple use cases, complexity increases with data volume, data complexity and the need for compliance.
If you need protect person data, intellectual property or just be sure that the answers of your chatbot are 100% correct, the effort to build this increases by factors.
Unless you use the thalabus platform that provides the right tooling for complex solutions.

thalabus facilitates the creation of a knowledge base, integrates data from diverse source systems, and automatically builds application-specific ontologies and graph databases. These also serve to compare, standardize, and ensure the quality of the imported data.

Based on this knowledge, thalabus-powered chatbots can generate accurate, verifiable answers in the desired language. While these answers are also generated using language models, they are based on the actual imported data and data from current websites. This virtually eliminates hallucinations and incorrect answers like those sometimes seen with ChatGPT and others. Because thalabus always maintains the entire chain of answer generation and source systems, the user can easily access the sources of information directly. This type of answer generation is an optimized, so-called Retrieval Augmented Generation (RAG).

Update February '24: Recently, even Gemini and ChatGPT offer a "Deep Research" mode for searching for answers on current websites. However, these models cannot access internal/sensitive company data.

Empower Your Customers with Self-Service and Support through AI Chatbots

halabus chatbots revolutionize customer self-service and support by providing instant answers and personalized assistance. While leveraging publicly available data for common queries, thalabus maintains the highest level of security for your sensitive customer data. Private information resides within your existing systems and is accessed only when specifically required, ensuring compliance and data protection.

With thalabus, access to sensitive data is strictly controlled through existing system authorizations. Users can even be guided through the process of requesting access if needed.

Connecting your systems to thalabus unlocks even greater potential, enabling chatbots to display dynamic graphs and visualizations directly from your data sources . Customers can effortlessly track orders, review purchase history, and even request changes (like delivery address updates) through natural language interactions – no need for them to understand the complexities of your backend systems.

thalabus can be seamlessly embedded within your existing portal and customized to reflect your brand identity.

thalabus provides the foundation for a truly effective self-service experience.

Read more Close

Knowledge Management and Semantic Search

t2b is working since 2012 in this field - mostly for Pharma companies. We help to extract and consolidate knowledge that is hidden in PDF-documents, the intranet and other sources.
"Classic" machine learning, manual management of ontologies and the generation of the RDF input for graph databases can now be supported and made more efficient by LLMs. But the biggest added value is on the query side: as the solution "understands" the concepts and related topics, it can provide answers to complex research questions.

When reading data, thalabus automatically builds application-specific ontologies and graph databases. These also serve to compare, standardize, and ensure the quality of the data. With LLM support, synonyms, homonyms, and other special cases are correctly classified/linked in the graph database. The various data and text blocks are also multiply indexed/tagged.

The semantic search then gives the user the ability to search using their own terms, even though different terms may have been used in the data sources. This "translation" works with languages, but also for users with different backgrounds: for example, an engineer/product developer will likely search for a product or concept using a different term than the PR manager.

The graph database is the foundation of thalabus, but the user can also jump directly to it with a single click and can research there directly via the GUI. However, we expect that natural language queries to thalabus will lead to the desired result so quickly that this option will rarely be used.

thalabus becomes the portal to your enterprise's knowledge

Knowledge Management

Until now, language models have rarely been used for search due to cost considerations. For data preparation, most systems also work with classic machine learning, algorithms, and classic programming.

thalabus has the advantage that questions are broken down into small elements, and the search across the vector/graph database is very efficient. The summarization then takes place in a cost-effective, local language model.

thalabus enables the creation of complex ontologies, providing the flexibility to query information from many different angles. This improves the classification, cleansing, and accessibility of an organization's knowledge.

Discover our comprehensive approach to data management and governance.Close

Data quality management

Are data from different sources always consistent? This is rarely guaranteed. When merging the data, thalabus will discover inconsistencies. We recommend parameterizing the system in such a way that it automatically corrects simple errors and passes difficult ones to a user workflow. In any case, it is always annotated and a list of the 'changes' is maintained, so that the differences from the source data can be clearly identified.

We recommend always correcting data errors in the source systems. Of course, thalabus can also take on this task or provide support in doing so.

For some organizations, the ability to ensure data quality will be a highly valuable, possibly even the most valuable, aspect of thalabus

Read more Close

Product Data and Catalogues

Your product data is widely distributed over systems, languages, media in very different quality levels? Your organisation is investing a lot of effort and money to integrate, validate, harmonize and then present the data?
The distribution to your partners, systems and catalogues is again quite some effort and requires a lot of error handling?
a thalabus based solution will bring quite some relief!

Read more Close

Scanning, OCR++, Document-Analysis

Paperless office? Digitalization? Many are still working on it. In fact, a lot of data is still on paper. To process it digitally, it needs to be scanned and digitized. Reading scanned images with classic OCR libraries works reasonably well. EXCEPT when the printing/copying quality is poor. This is where AI comes in, just like when processing the generated character sequences. AI can also support problematic cases such as the correct sequencing of multi-column texts, image and table processing. Thanks to a specialized partner, we can offer this as well

Read more Close

AI support for Enterprise Architecture

Enterprise architecture management is applied best practices in many enterprises. t2b is supporting medium to large companies and governnment units in this domain since 2001. AI support is making the architects' work on the data gathering, creation and consolidation easier.
But the real added value of AI comes by making the access to the architecture information easily accessible for any stakeholder.

With thalabus's pre-configuration for Enterprise Architecture Management, organizations can immediately leverage built-in ontologies from TOGAF and Archimate, and easily incorporate their own. This makes data quality assurance during the import process a straightforward task.

EAM-System and its surrounding systems

By integrating various systems within the EAM system across the entire lifecycle, a holistic view of projects, systems, and processes is possible. thalabus reads the data from the EAM system (e.g., Mega), the project management system (e.g., Jira), the version control system (e.g., Git), and the various document systems (e.g., Confluence) and brings them together.

This substantially reduces the effort required for data maintenance and queries.

thalabus not only creates answers from its local data copy, but also provides direct access to functions, graphics, and tables within the integrated systems. This makes finding the right source system and the specific report you need within it much easier.

thalabus streamlines access to all architectural artifacts, becoming the go-to portal for the entire enterprise.

Data quality management

We recommend always correcting data errors in the source systems. Of course, thalabus can also take on this task or provide support in doing so.

For some organizations, the ability to ensure data quality will be a highly valuable, possibly even the most valuable, aspect of thalabus

Read more Close

Vision. Plan. Success.

We bring you to the destination

Our goal is to provide you with the tools, strategies and resources you need to realize your vision and lead your business on a fulfilling and profitable journey.

AI-supported Solutions

Your Data. Your AI.

Secure. Data stays in your data center.
Efficient. thalabus has low infrastructure requirements.
Fast. AI solutions are created in a few weeks on thalabus.
Correct. Your answers are based on your data and the source can be displayed.

Chatbots for Customers and Employees

Knowledge Management and Semantic Search

Product Data and Catalogues

Scanning, OCR++, Document-Analysis

AI support for Enterprise Architecture

Vision. Plan. Success.

We bring you to the destination

News

Links

12.Nov'19: we are ISO9001 certified!

AI-supported Solutions

Your Data. Your AI. Secure. Data stays in your data center.Efficient. thalabus has low infrastructure requirements.Fast. AI solutions are created in a few weeks on thalabus. Correct. Your answers are based on your data and the source can be displayed.

Chatbots for Customers and Employees

Knowledge Management and Semantic Search

Product Data and Catalogues

Scanning, OCR++, Document-Analysis

AI support for Enterprise Architecture

Vision. Plan. Success.

We bring you to the destination

News

Links

12.Nov'19: we are ISO9001 certified!

Your Data. Your AI.

Secure. Data stays in your data center.
Efficient. thalabus has low infrastructure requirements.
Fast. AI solutions are created in a few weeks on thalabus.
Correct. Your answers are based on your data and the source can be displayed.