®KINEVA® / MODELLSERIE
Answers about your model and process data.
KINEVA® One is the series of large language models within the KINEVA® model family. It makes results from KINEVA® Smart City and KINEVA® HORIZON understandable and connects them with other process data and internal knowledge. Ask questions in natural language, including about relationships that previously required complex analysis.
Discuss KINEVA®KINEVA® / MODELLE
Visual models capture. KINEVA® One explains.
KINEVA® Smart City
The models detect road damage, illegal dumping, road and traffic signs, and roadwork delineators. They capture where roadworks begin and end, the positioning and condition of traffic signs, and changes to infrastructure. GUSTAV processes camera images directly on site.
KINEVA® One combines these findings with existing asset records and maintenance information. Teams can assess where traffic signs are missing, positioned differently, or need cleaning, repair or replacement. HORIZON Smart City forwards the results to the responsible teams.
KINEVA® HORIZON
The visual industrial models inspect copper plates, metal and sheet metal for surface defects and detect foreign objects in material streams. KINEVA® One answers questions about these results and the associated process data.
KINEVA® One answers your questions
The large language models connect detection results with process data and internal knowledge. You can ask where large amounts of bulky waste were detected or how surface defects have changed over time. KINEVA® One makes these relationships accessible in natural language.
10-year lifecycle support for fine-tuning and optimization of our model families.
KINEVA® ONE + HORIZON
REBOTNIX® models detect. KINEVA® One answers. HORIZON forwards.
How they work together.
GUSTAV processes camera data on site. REBOTNIX® visual models deliver detection results. KINEVA® One answers questions about the combined data and draws on internal knowledge. HORIZON Smart City and HORIZON Industrial deliver information into existing workflows.
Explore HORIZONKINEVA® ONE / RACK-SYSTEM
Fits on a desk. Scales to a data center.
KINEVA® One is available as a compact rack system that fits on a desk. Companies can test their own AI stack under real conditions before deciding on larger infrastructure investments. As requirements grow, the solution scales from a single rack to a data center. Operated by REBOTNIX® or on your premises on request; no data is shared with third parties.
INSIDE THE RACK
Depending on the configuration, the rack uses four NVIDIA GB10 or four NVIDIA Jetson AGX Thor (T5000) modules. Each module has 128 GB of unified system memory for CPU and GPU. The Thor configuration also includes PVA 3.0 for vision processing; REBOTNIX® tunes the software for the selected hardware.
IN THE DATA CENTER
For larger installations, REBOTNIX® runs KINEVA® One in the data center on systems with NVIDIA RTX 6000 graphics cards. The same software foundation connects results from GUSTAV edge devices, the HORIZON platforms and orchestration by KINEVA® One.
REBOTNIX® is an NVIDIA Preferred Partner.
For technical teams
How KINEVA® One handles large data volumes.
Physical AI
CNN models detect objects in camera images. PCNN models combine image and sensor data. Suitable models run on GUSTAV directly at the point of use.
Language and image
LLMs and VLLMs help interpret text and images. Internal RAG retrieves relevant information from your own knowledge sources. Accelerated processing and cost-conscious tools make large streams of findings usable.
Developed in-house
The KINEVA® Training Suite is our software for training the models. REBOTNIX® maintains and takes responsibility for the model families and develops them for specific applications.
Explore technical detailsWatchdog LLM engine and tools
At its core is a Watchdog LLM engine optimized by REBOTNIX®. Its tools let language models use detection results, process data and knowledge from the internal RAG in a targeted way and return answers to HORIZON. REBOTNIX® develops and maintains this technical layer itself.
Your application
Where do you need a clearer picture?
Show us which findings and internal information need to come together. We can discuss how KINEVA® One, GUSTAV and HORIZON work together, and which size and operating location fit.
Discuss your application