Private AI, hosted in Ireland.
Run open-source AI privately on your organisation’s data using tandemcompute infrastructure in Ireland.
Use AI on company data without sending it to a public AI platform.
A private endpoint, your documents, your users.
The model
- Selected open-weight model
- Private model endpoint
- Dedicated capacity
Your data
- Retrieval over your documents (RAG)
- Document storage
- Monitoring and support
Access
- Authentication
- Private API
- Optional tandemcompute web interface
Organisations whose data cannot leave their control.
Regulated
- Law firms
- Financial services
- Healthcare
- Government suppliers
- Regulated companies
Sensitive by nature
- Professional services
- Enterprise
- SMEs with sensitive data
Not “which model do you want?”
We start with what you are trying to make AI do for your organisation. The model, the retrieval layer and the interface follow from that.
Organisations whose data cannot leave their control.
Weeks: a scoping call, then setup on dedicated capacity.
Dedicated capacity in Ireland. Your documents and prompts never share hardware with another customer.
A named tandem engineer, by email and phone, Irish hours with out-of-hours cover.
The questions that usually come first.
Where exactly is our data?
On a dedicated node at a named tandem site in Ireland, encrypted at rest and in transit, on infrastructure tandem owns and operates.
Who at tandem can see it?
Nobody, in the ordinary course. Engineers see operational metrics (load, latency, health), not content. Access to your environment requires your written authorisation and is logged.
Does the model learn from our data?
No. Nothing you send is used to train or fine-tune any model unless you commission a fine-tune on your own data, which stays yours.
Can it answer from our own documents?
Yes. Retrieval over your documents is the usual configuration: the model answers with citations back to your files.
Which model will we get?
An open-weight model chosen for your documents, languages and volume during scoping. You can change it later without changing anything else.
What happens if we leave?
You export your data and any fine-tuned weights; we decommission the endpoint, wipe the storage and confirm it in writing. No lock-in beyond the monthly term.
Other ways to work with tandemcompute.
Need GPU compute?
You already know you need accelerated infrastructure.
gpu compute → modelsNeed an AI model?
Open models through an API, without managing the infrastructure.
tandemcompute models → workloadsHave a GPU-heavy workload?
You care about getting a task completed, not managing hardware.
workloads →