Your clients' call data is the product.
It shouldn't also be someone else's training data.
Private LLM hosting on a dedicated machine in EU-Central, for call centers and BPOs running transcript summarization, QA scoring, or agent assist on client call data without routing it through a third-party AI vendor your client never approved.
Not legal advice. This page describes GPUwerk's infrastructure and contractual commitments. It is not an opinion on your obligations to the clients whose calls you handle, or on whether a given setup satisfies a specific client contract. Check with your own legal team before deciding what can run where.
Where an LLM fits into a call center or BPO operation
Call center work generates a constant stream of client call data, and most of it is a candidate for automation that still needs somewhere safe to run.
Transcript summarization
Turning a raw call or chat transcript into a short summary for the next agent, a supervisor, or the client's own reporting. The transcript itself usually names the caller and describes their account, order, or issue in detail, which is exactly the kind of data your client's own contract likely restricts.
QA scoring
Scoring a sample of calls against a quality rubric, or flagging calls for a human reviewer to check. Running this at volume through a shared API means every scored transcript, across every client account you handle, passes through one vendor's infrastructure.
Agent assist
Suggesting a next-best response or pulling up relevant knowledge base content while an agent is live on a call. Latency and data handling both matter here: a dedicated instance close to your call infrastructure avoids both the round trip to a shared API and the question of where that live call content ends up. See the private AI support ticket triage guide →
Multi-client operations and data segregation
A BPO typically handles calls for several clients at once, each under a separate contract with its own data handling terms. A GPUwerk instance is dedicated to your business, not automatically partitioned by client, so if one client's contract requires that its data never touch infrastructure shared with another client's workload, plan for separate instances per client rather than assuming a single deployment covers every contract. That's a decision for your own client management and legal team, not something GPUwerk can determine on your behalf.
What's actually in the contract
For your operations lead or IT lead to review directly.
| Question | Answer |
|---|---|
| Where does it run? | EU-Central, on a dedicated single-tenant machine assigned to your business |
| Who operates it? | PRINT IT! SE, a Societas Europaea registered in Prague, Czech Republic. No US parent entity. |
| Who can reach the instance? | Through the instance itself, only holders of your SSH keys; password login is disabled fleet-wide. GPUwerk keeps infrastructure administrator access to the underlying machine, as on any hosted service, and under the DPA does not use it on your content except at your request for support or where a legal obligation requires it. |
| Does GPUwerk see call transcripts or scores? | No. We host the hardware and, under the DPA, do not access, read, copy, index or analyse workload content. |
| Sub-processors for the workload? | None, listed at /legal/sub-processors |
| DPA (GDPR Art. 28)? | Published at /legal/dpa, no charge |
| Multi-client isolation? | One instance per business by default. Separate instances per client if your client contracts require it; this is your call to make. |
| Data on termination? | Container and workspace volume deleted from the node, then the node is sanitised before reassignment. Filesystem deletion, not a cryptographic erase; export what you need before terminating, since it isn't reversible. |
| Pricing | Spark-1x from $0.79/hour on demand, billed per minute; Spark-2x from $1.79/hour. A customer-requested stop holds the reservation at 75% of the running rate. See /pricing for full detail. |
Questions we get from call center and BPO teams
We handle calls for multiple clients. Can one dedicated instance serve all of them?
Technically yes, but consider your client contracts first. Some clients require their data to stay on infrastructure they've separately approved. A single GPUwerk instance is dedicated to your business, not automatically segmented by client, so if a contract requires per-client isolation you would run separate instances. GPUwerk does not access, read, copy, index or analyse workload content except at your request for support or where a legal obligation requires it.
Does GPUwerk see the call transcripts or QA scores we process?
No. GPUwerk operates infrastructure, not models. Transcripts, QA scoring data, and agent assist output processed on your instance are handled by software you install, and under our data processing agreement GPUwerk does not access that content except at your request for support or where legally required.
Do you sign a DPA covering client call data we process on behalf of our own customers?
Yes, a standard GDPR Article 28 DPA is published at /legal/dpa at no charge, and there are no sub-processors for instance workloads. Whether that satisfies your own downstream contractual obligations to your clients is for your legal team to confirm.
Does GPUwerk train on our data?
No. GPUwerk runs infrastructure, not models. There is nothing for GPUwerk to train on, since we do not access instance content except as described in the DPA.
Related reading
Private AI for customer support · Private AI for support ticket triage · Private LLM hosting for staffing agencies
Related pages
Keep client call data off infrastructure they never approved.
Talk to the people who run the racks, or start with a pilot and a practical rollout plan.
Talk to us Deploy an instance