Swiss AI API: Swiss-Hosted, Zero Data Retention

Journey of a request

Follow one request from your app to a Swiss GPU and back

Press play, or step through it. Then flip Swiss-only mode and simulate an outage to see the difference: a detour abroad, or an honest error.

Your appAnywhereTLS
API gatewayCloudflare network·
GPUSwitzerland·
Failover GPU poolOutside Switzerlandnever used
Answer returnedBack to your app·
Content discardedNowhere·

Your app sends text and questions over TLS. Everything on the wire is encrypted.

response headers appear when the answer comes back…

content stored: 0 bytes

Swiss-only mode ON: the failover pool is locked out. An outage returns an error instead of a detour.

The facts

Swiss hosted AI, in six plain facts

Where the GPUs are, what is kept, and what never is. Each card is the exact wording we stand behind.

GPUs in Switzerland
Our primary inference pool runs on dedicated GPUs located in Switzerland. Every API response tells you where it was processed in the x-laya-region header.
Swiss-only mode
One switch per workspace (or the x-laya-residency: ch header per request) and requests are only ever answered in Switzerland. If the Swiss pool is unavailable you get an error, never a silent detour abroad. It is an add-on: +15% credits per request, still about 20% below Jev.
Zero content retention
The text and questions you send are processed in memory and discarded when the answer is returned. They are never written to a database or log, and never used to train anything. One exception to know about: the optional chat agent (POST /v1/ask, the MCP ask tool, dashboard Chat) first passes your prompt to a third-party language model to plan the questions, in the US by default and in Switzerland under Swiss-only mode.
Account data in Zurich
Accounts, API keys (stored only as SHA-256 hashes), credit balances and usage metadata live in a Postgres database in the AWS Zurich region (eu-central-2).
Minimal metadata, 30 days
For billing and debugging we keep request metadata only — time, status, number of questions, latency — for 30 days. Daily totals are kept for invoicing.
Encrypted end to end in transit
All traffic uses TLS. The API gateway runs on Cloudflare’s network, which forwards requests encrypted and does not store their content; the GPU servers accept requests only from our gateway.

Zero data retention

What stays and what’s gone

Hover or tap each kind of data to see where it lives and for how long. The short version: your content is gone the moment you have your answer.

Where each kind of data lives and for how long. Hover or focus a row for details.
Data	Where it lives	How long
Your text	GPU memory (RAM) only	Gone
Your questions	GPU memory (RAM) only	Gone
Your API key	Zurich database, as a SHA-256 hash only	The hash, while the key exists
Request metadata	Zurich database	30 days
Daily usage totals	Zurich database	Kept for invoicing

Medical & sensitive data

Built for sensitive data

No retention plus Swiss processing makes Laya Studio a fit for sensitive text such as patient messages, clinical intake notes, HR cases or financial correspondence — the kind of data that should not sit in a US AI provider’s logs.

Pick a use case

Clinical intake routingHR and legal case sorting

Synthetic message (fake data, edit freely)

Please don’t paste real patient data into a public demo.255/600

3 questions · nothing stored

Answers

route

Which team should handle this patient message first?

same_day

Does the patient describe symptoms that need same-day care?

side_effect

Does the message mention a possible side effect of a medicine?

Press “Run in Switzerland” to get a probability for every question.

This demo always runs in Switzerland and shows the region the answer came from. More ideas on the use-case pages.

For developers

Enforce Swiss-only processing in one switch or one header

Swiss-only mode can be set for a whole workspace or for a single request. A request can tighten residency, never loosen it.

For the whole workspace
Open Dashboard → Settings → Data residency and turn on Swiss-only. Every key in the workspace is then answered in Switzerland only.
Per request
Send the header x-laya-residency: ch. If the Swiss pool is unavailable you get HTTP 503 residency_unavailable, never a detour.
Verify every response
Every response carries x-laya-region (where it ran) and x-laya-residency (ch or any). Log them, or assert on them.
Read the API docs →

swiss-only.shbash

curl -i https://api.laya.studio/v1/systemone \
  -H "Authorization: Bearer $LAYA_KEY" \
  -H "x-laya-residency: ch" \
  -H "content-type: application/json" \
  -d '{"state":"Synthetic patient note: rash after new cream.",
       "questions":{"side_effect":{"type":"noul",
         "instructions":"Does it mention a possible side effect?"}}}'

# Response headers
# x-laya-residency: ch
# x-laya-region: ch

nFADP & GDPR

nFADP, GDPR and your DPA: what we say, and what we don’t

Swiss data protection, stated carefully. We only claim what we can back up.

Frameworks

Designed to support compliance with the Swiss Federal Act on Data Protection (nFADP, in force since 1 September 2023) and the EU GDPR.

Data Processing Agreement

Our Data Processing Agreement (laya.studio/dpa) is part of the Terms and applies automatically when you send personal data; a signed copy is available on request: privacy@laya.studio.

Read the DPA →

Please read: the caveat

You remain responsible for your legal basis to process personal and health data. Laya Studio holds no formal certification (such as ISO 27001) today and does not sign HIPAA BAAs.

Details in our privacy policy and on the security page.

FAQ

Swiss data residency: common questions

Where are the GPUs that process my requests?

Our primary inference pool runs on dedicated GPUs located in Switzerland. Every API response tells you where it was processed in the x-laya-region header. Accounts, hashed API keys, credit balances and usage metadata live in a Postgres database in the AWS Zurich region (eu-central-2).

Is anything I send logged or stored?

No content. The text and questions you send are processed in memory and discarded when the answer is returned. They are never written to a database or log, and never used to train anything. One exception to know about: the optional chat agent (POST /v1/ask, the MCP ask tool, dashboard Chat) first passes your prompt to a third-party language model to plan the questions, in the US by default and in Switzerland under Swiss-only mode. For billing and debugging we keep request metadata only — time, status, number of questions, latency — for 30 days. Daily totals are kept for invoicing.

Can I use the API for medical or patient data?

No retention plus Swiss processing makes Laya Studio a fit for sensitive text such as patient messages, clinical intake notes, HR cases or financial correspondence — the kind of data that should not sit in a US AI provider’s logs. You remain responsible for your legal basis to process personal and health data. Laya Studio holds no formal certification (such as ISO 27001) today and does not sign HIPAA BAAs.

Is my data exposed to the US CLOUD Act?

Here is exactly what touches what. The API gateway runs on Cloudflare, a US company: it forwards your requests encrypted over TLS and does not store their content. Inference runs on GPUs in Switzerland, and the account database is in Switzerland (the AWS Zurich region, eu-central-2, operated by Amazon). Because the content of your requests is never written to a database or log, it is not part of any stored data; account data and request metadata are stored, in Zurich. Other US-headquartered providers do touch some data: Stripe (payments), Resend (account emails), Google (only if you sign in with Google), and, outside Swiss-only mode, a US-hosted language model that plans chat-agent prompts and an optional cloud GPU failover. Swiss-only mode keeps both of those in Switzerland. The full list is on our subprocessors page. Whether that is enough for your risk assessment is your call; we are happy to answer detailed questions at privacy@laya.studio.

What happens if the Swiss GPUs are down?

Without Swiss-only mode, the gateway may fail over to a non-Swiss GPU pool so your request still gets an answer, and the x-laya-region header shows where it ran. With Swiss-only mode on (Dashboard → Settings → Data residency, or the x-laya-residency: ch header), a request is only ever answered in Switzerland: if the Swiss pool is unavailable you get an HTTP 503 “residency_unavailable” error, never a silent detour abroad.

Do you sign a Data Processing Agreement (DPA)?

Our Data Processing Agreement (laya.studio/dpa) is part of the Terms and applies automatically when you send personal data; a signed copy is available on request: privacy@laya.studio. Designed to support compliance with the Swiss Federal Act on Data Protection (nFADP, in force since 1 September 2023) and the EU GDPR.

Is my data used to train models?

No. The text and questions you send are processed in memory, discarded when the answer is returned, and never used to train anything.

How can I prove where a request was processed?

Every API response carries two headers: x-laya-region (where the request was processed, “ch” for Switzerland) and x-laya-residency (“ch” when Swiss-only mode applied to the request, otherwise “any”). You can log and assert both in your own code.

Start with 5 free runs, answered in Switzerland

Then $0.0294 per 1M input tokens — 30% below Jev. Create a key, turn on Swiss-only mode, and check the x-laya-region header on your first response.

Get a free API key See pricing

Swiss-hosted AI decisions. Nothing you send is ever stored.

Follow one request from your app to a Swiss GPU and back

Swiss hosted AI, in six plain facts

GPUs in Switzerland

Swiss-only mode

Zero content retention

Account data in Zurich

Minimal metadata, 30 days

Encrypted end to end in transit