Join the project: SMEs and researchers welcome
Eleven partners form the first cohort — and access expands as delivery progresses. Whether you are a Welsh business seeking operational productivity gains or an academic team seeking compute collaboration and joint grant bidding across Wales and beyond, you are welcome to apply.
Model service access
Connect your tools and workflows to Wales-hosted open-weight models through a standard API or a managed workspace — the lightest-touch way to try private AI.
Best for: software teams, data-savvy SMEs, anyone curious what "the same tool but local" feels like.
Structured partner project
A scoped collaboration with a Bangor academic lead: a defined productivity goal, a secure workspace, fine-tuning or RAG work, and measurement against a frozen baseline.
Best for: SMEs with a specific process they want to transform — documents, vision, forecasting, agents.
Edge AI deployment trial
Take one of four DGX Spark systems to your site for a managed trial — local inference, offline operation and low-latency testing, with the unit returning to Bangor afterwards.
Best for: care settings, utilities, environmental monitoring, manufacturing and remote sites.
Academic research partnership
Partner on frontier AI research: multi-GPU model fine-tuning (LoRA/QLoRA), autonomous agent systems, PhD/postdoc compute quotas, and joint UKRI/EPSRC grant proposals across Wales and beyond.
Best for: university academics, doctoral researchers, research institutes, and grant consortia.
From first email to measurable result
Projects are selected on readiness, technical fit and capacity. There is no charge to express interest, and no commitment until agreements are in place.
Start with one email
Write to [email protected] or use our enquiry form.
Express interest
Tell us who you are and the outcome you want. Ten minutes of writing is plenty.
Scoping conversation
We test feasibility together: data, systems, success metrics and effort on your side.
Agree & onboard
Data, IP, security and equipment-loan agreements protect everyone. Your workspace is provisioned; training is arranged.
Deliver & measure
Build, trial (central or on-site with an edge unit), and measure against the baseline you helped define.
Share the story
With your permission, results become an anonymised case study — proof that Welsh SMEs can do this.
What partners receive — and contribute
You receive
- Access to enterprise AI compute with no capital purchase
- Wales-hosted open-weight models — no commercial per-token charges during the pilot*
- A secure, isolated project workspace with training and technical support
- A named Bangor academic lead — plus student project teams where useful
- A managed edge AI trial at your premises, if your case needs one
- Reusable workflows and evidence of productivity gains for your own business case
*Subject to capacity and governance requirements.
You contribute
- A clearly defined challenge with a measurable productivity outcome
- Appropriate access to data and domain expertise
- Staff time for scoping, testing and evaluation
- Site access for trials where edge systems are used
- Honest feedback and measurement — that's how the whole programme earns its keep
Who can take part?
The programme is built for Welsh businesses, with North Wales at its heart — the first cohort spans software, healthcare, environment, professional services, translation and education, from Anglesey to Flintshire. New SMEs are welcome to express interest in joining, subject to the project's eligibility and participation arrangements; access expands during delivery as capacity allows.
Each selected project receives a secure workspace, training and technical support. Data, IP, security and equipment-loan agreements protect all parties.
Do we need AI expertise in-house?
No. Academic leads and student teams carry the technical heavy lifting — you bring the problem, the data and the judgement. Curiosity helps more than code.
What does it cost us?
Your staff time, data access and evaluation effort. Infrastructure access, training and technical support come with the project; there are no commercial per-token charges during the pilot, subject to capacity and governance.
What happens to our confidential data?
It works inside an isolated project workspace hosted at Bangor University, governed by project-specific controls designed to protect confidentiality and prevent your data training external third-party models. Formal data, IP and security agreements are signed before work starts.
Can we keep the edge AI unit?
The four DGX Spark systems run on managed, temporary rolling deployments — in and out of partner sites as trials need, returning to Bangor for reconfiguration. All grant-funded assets return to the central pool by 31 March 2028.
Which models will we get?
Managed open-weight models such as Qwen, GLM and DeepSeek — or suitable successors chosen on performance. The field moves fast; the platform is specified so the platform can always host the best available.
When can projects start?
The service is expected to go live within three months of equipment commissioning (delivery completes January 2027). Scoping conversations are open now — early interest shapes the queue.
Ready? One email starts it.
Tell us about a task where AI could save hours, unlock capacity or win work.