Business reporting with everyday language

Analyze data quickly and intuitively – simply by natural language input. Our generative AI generates complete reports and makes business intelligence available to everyone.

Security

Our language models can be operated locally, so your data remains secure.

Connection

We generate all results from raw data from a wide variety of source systems.

Speed

All company data is just one single question away.

Intuitivity

Our digital assistant frees users from all technical barriers!

Radically simplify reporting

We reduce manual work from hours to seconds

Autogenerated Dashboards

Our AI creates complete analyzes & dashboards, tailored to every question.

Augmented Analytics

Analysis results are automatically enriched with valuable information.

Insights & Summaries

Auto-generated insights for any query are summarized in summaries.

Next Generation BI

We analyze our sales data with the language assistant from semantic one. The speed at which we gain transparency about our data without training and without prior knowledge is impressive

Tim Klauke
CEO

The advantage of semantic one is that entire work processes can be automated, which with the previously known way of working could only be carried out through many manual individual steps and a lot of time.

Christian Schmidt
Leader Controlling

What we do

Our Technology

Our technology was specifically developed for analyzing structured data and delivers accurate results.

Our Solution

We provide digital assistants for your business reporting – via our cloud or on-premise in your company.

Our Services

We build tailor-made solutions for your business reporting and reflect your company’s individual language use in our language models.

FAQ

Frequently asked questions

Natural language control means automation. Work steps that were previously carried out manually are completed within seconds by a digital assistant. When creating reports and dashboards, there are a variety of repetitive tasks that can be easily automated. Gen AI is capable of generating complete reports, which requires hours or even days of manual effort with mouse/keyboard-based interfaces.

Our generative AI was developed specifically for use in reporting and is not based on a classic LLM architecture. Language models from semantic one are 99 percent auto-generated and also contain master and meta data from a reporting cube. These language models are relatively small, so they can also be operated offline in local installations.

When auto-generating the language model, the master data can either be read directly from a data warehouse or generated from the transaction data. Using a help assistant, users can view both the data model and the master data and select them directly when formulating the question.

Over time, every company forms its own language world, which uses very different terms or term meanings from case to case. In the language models from semantic one you can add synonyms, short names, abbreviations and special terms of all kinds and thus reflect the company’s individual language world.

Semantic one works with a so-called ‚white box‘ solution. The core of our NLP technology does not consist of neural networks, but rather an open NLP algorithm. At every level of semantic recognition we can add logic routines and thus ensure the quality of our results. Our algorithms are highly tailored to work on structured data and deliver reliable results.

The ability of so-called LLMs (Large Language Models) to access structured knowledge and process it precisely is limited, so their performance lags behind that of task-specific architectures. The avoidance of inconsistencies and hallucination effects in LLMs is currently still the subject of scientific research. An example of this are Retrieval-Augmented Generation (RAG) models that combine pre-trained parametric and non-parametric memories for language generation, thereby improving the quality of results. To date, efforts in this regard have not yet been able to achieve relevant market maturity.

Yes. Our reporting solution can be operated either SaaS-based or in a local in-house cloud. Since we generate reporting-specific, local language models, our solution does not require external Internet access, but can be operated within a company network as a local in-house installation.

If you want to query a database with semantic one’s reporting assistant, you must provide access to the data or implement the solution as an in-house cloud installation in your company. The language model for the respective data model is auto-generated based on the movement data. As soon as the language model is ready, the database can be queried. We offer all interested parties a 3-month, free test phase.

With semantic one’s voice assistant, users create analyzes and dashboards – simply by voice. No training or induction period is provided. A help assistant provides the data model and access to the master data, as well as assistance in formulating questions.

Yes. We offer all interested parties a free trial period of 3 months. During this time, our solution can also be installed and tested in a live situation with customers. The test phase can be canceled within 3 months without incurring any costs.

Yes. You can build standard reports for our assistant that can be accessed with a mouse click. We also offer the option of generating individual dashboards based on predefined layouts – simply by voice input.

In principle, it is possible to connect any number of data sources, from different databases, source systems or from a data warehouse. If source systems make their data available via API, a connection can also be made this way.