What is the BDSi Local AI cluster?
The BMS Lab now offers six high-performance AI workstations to support and accelerate AI research. These workstations are based on an AMD Strix Halo 395+ AI Pro chipset, making them ideal for experimenting with small- to medium-sized local AI models.
These Framework PCs can support a wide range of research applications, for example exploring AI-assisted interventions, analysing qualitative data, developing conversational agents, or testing human–AI interaction scenarios, as demonstrated in this research project: BuddyGPT.
Several robust multi-purpose models are availeble, such as Google's Gemma 4 and Qwen 2.6. Depending on your needs and available capacity, different models can be made available on request. This covers a large and growing list of open-weights models from frontier AI labs, as well as specialized models from leading international research teams.
How can the BDSi Local AI cluster help me?
We offer three distinct services backed by the BDSi Local AI cluster. From a simple chat interface for conversational AI with a local LLM, to bare-metal access for prototype research applications using a tailored stack of local AI tools.
Chat UI Access
Use a browser-based chat interface to interact directly with available AI models in a conversational style.
Best suited for tasks such as:
- General AI experimentation
- Exploring model capabilities without technical setup
- Ideation and refinement of research proposals
- Anything you might use ChatGPT for, but prefer to keep local
API Access
Access models programmatically through API endpoints.
Best suited for tasks such as:
- Replicable qualitative analyses (e.g., summarizing conversations, topic modelling, classification, etc.)
- Creating AI-powered agents (e.g., chat agents (furhat, chatbuddy), ai-assisted web scraping, etc.)
- As a local alternative to commercial cloud hosted models for 3rd party tools (e.g., coding agents)
Bare metal access
Researchers can request exclusive remote access to a dedicated Framework PC for implementing and testing more advanced AI workflows and software stacks. For projects with complex security requirements, the server can be physically isolated for truly local and off-line AI in a secure environment under your control.
Best suited for:
- Complex applications that require multiple interacting services (e.g., a prototype research application requiring a large language model, database, embedding model, customized sub-agents, MCP tools, and a web server)
Best practices
Accessing and using the BDSi Local AI cluster
To gain access to the cluster, please register a request, describe your use case, and choose your desired access level from the available options. Our team will assess your request, and contact you about the next steps depending on your needs.
Asking for help and reporting issues
For support, questions, or suggestions about the BDSi Local AI cluster, contact bdsi@utwente.nl.
You are responsible for your data
Make sure you are following the BMS and UT AI use guidelines and policies
Further support
A guide describing how to use the Chat UI will be available soon. For further support, or questions about the API or bare-metal access, contact bdsi@utwente.nl.
