On this page
AI for the financial sector
Financial institutions operate under strict rules around data, operational resilience and accountability. HostYourAI runs open models on GPUs in European datacenters via vLLM, so your data stays in the EU and you keep control of the whole chain. No vendor lock-in, because the API is OpenAI-compatible and you can always leave.
Why financial teams host their AI in the EU
Data residency and operational resilience
All processing happens inside the EU, with no US cloud provider in the chain. That aligns with the thinking behind DORA around control of your ICT chain and reducing dependency on a single party outside Europe.
Auditable and accountable
You run open weights with reproducible behaviour, and you see what happens per request. That makes internal control, model validation and accountability towards supervisors easier than with a closed black box.
A Data Processing Agreement and no training on your data
A Data Processing Agreement (DPA) is available and the subprocessor list is public. Your prompts and documents are not used to train models.
What you use AI for in finance
Think of summarising reports and regulation, analysing documents in KYC and AML processes, answering internal policy questions over your own documentation and drafting communication. Decisions stay with your team: the model speeds up the groundwork and delivers a draft that a human reviews. Do not use it for autonomous investment or credit decisions.
Finance requirements and how HostYourAI meets them
| Requirement in finance | How HostYourAI meets it |
|---|---|
| Data residency inside the EU | European GPUs, no US cloud in the chain |
| Control of the ICT chain (DORA thinking) | Open models, public subprocessor list, no lock-in |
| Accountability and model validation | Reproducible behaviour, per-request logging |
| Protection of customer and transaction data | Processing in the EU with a Data Processing Agreement |
| Isolation for sensitive workloads | Dedicated GPU instances alongside the shared Router |
| Predictable cost | Pay-as-you-go per token or per hour, visible per request |
Formal frameworks such as DORA and supervisory requirements sit within your own governance. We provide the technical building blocks that fit: EU processing, a DPA, isolation, transparency and auditability.
How to start
Link a knowledge base with your own policies and documentation; the Router automatically injects the relevant context, so answers are grounded in your sources. Start on the shared Router and scale to a dedicated instance once you move to production or more sensitive data.
from openai import OpenAI
client = OpenAI(
base_url="https://hostyourai.com/api/v1",
api_key="hyai-...",
)
resp = client.chat.completions.create(
model="llama-3.3-70b",
messages=[{"role": "user", "content": "Summarise this regulation."}],
extra_body={"knowledge_base_id": 42},
)
print(resp.choices[0].message.content)Want to know more about linking knowledge? Read the guide Build RAG on EU GPUs.
Questions about AI in finance
Does our data stay inside the EU?
Yes. Models run on GPUs in European datacenters and your prompts and outputs never leave the EU. There is no US cloud provider in between.
Does this help with DORA?
We provide building blocks that fit the DORA thinking around control of your ICT chain: EU processing, a public subprocessor list, no lock-in and per-request logging. The formal DORA track runs through your own governance.
Can I use it for KYC and AML?
Yes, as support: summarising documents and making them searchable over your own sources. A human reviews and decides; the model takes no independent decisions.
Is our data used to train models?
No. There is no training on your input or output, and open weights do not learn from your traffic.
Am I locked in?
No. The API is OpenAI-compatible, so you point your existing client at our Router and can always leave. No vendor lock-in.
Can I get isolation for sensitive workloads?
Yes. Alongside the shared Router you can run a dedicated GPU instance for isolation and predictable performance.