Trust & Security

Your patent data stays yours.

Patent sets, product documents and invention disclosures are some of the most sensitive files a company holds. Here is how Semantyr handles them, in plain words.

Your data is never used to train any model. Every workspace is isolated from every other. You work through defined workflows, not an open chat box.

Talk to us about your requirements Email a security question
Built on trust

Four areas, one promise

Reliable and repeatable also means safe to put your most sensitive patent work into.

Infrastructure

Hosted on Google Cloud Platform (GCP), on Kubernetes. Cloud-native and built to scale with your portfolio.

Privacy

No open prompting. Every interaction runs through a controlled workflow. Data is isolated between workspaces, and a privacy policy is in place.

Training

Customer data is never used for model training. The analysis is done by a mathematical model, not a model trained on world data.

Security

We use GCP's infrastructure-level security, encryption, access control and isolation. On top of that we keep workspace-level data separation and privacy controls in the application.

TRUST
How your data moves

The math analyses. The language model only writes.

A mathematical model does the grouping, counting and measuring. A language model sees the measured results at the final step and puts them into plain sentences.

Your uploadPatent set, product document or disclosure
Isolated workspaceOn Google Cloud Platform
Mathematical modelGroups, counts and measures
Language modelWrites the results in plain words
Your reportEvery number traces to its patents

At no step is your data used to train a model. The analyst stays in control throughout.

Why Google Cloud

We chose infrastructure that has already earned trust

Your patent data deserves the same foundation that large organizations rely on. Google Cloud Platform follows many of the most respected security and privacy standards. That is why we chose it to keep your data safe. Google has chosen to work with us too, through its Cloud Program for startups.

Google Cloud Platform follows
ISO 27001ISO 27017ISO 27018ISO 27701SOC 2SOC 3GDPR-aligned data protection practices

Encryption, access control and isolation come from this infrastructure. Semantyr adds workspace-level separation and privacy controls on top.

Google for Startups

Chosen for the Google for Startups Cloud Program

Semantyr and the Google for Startups Cloud Program
Enterprise direction

If your organization has its own requirements

Larger teams often need more than the standard setup. Tell us what yours needs and we will tell you honestly what we can do today.

Regional hosting

Host your data in a region you choose, for example in Europe, on a Custom subscription.

Custom plan

Your own model

Use a private, sandboxed model that you already have, instead of a public one. Private and offline deployments can be discussed, depending on your needs.

Custom plan

Your templates and style

Configure templates and report style so projects, reports and opinions follow your standards.

Custom plan

Access controls

Enterprise deployment and access-control capabilities are being expanded step by step.

In progress

Questions from procurement

Do you train any model on our data?+
No. Customer data is never used for model training. The analysis itself is done by a mathematical model, not a model trained on world data.
Where is our data stored?+
On Google Cloud Platform infrastructure, in an isolated workspace. Regional hosting, for example in Europe, is available on a Custom subscription. Tell us what you need and we will arrange it.
Which security standards apply to our data?+
Your data is hosted on Google Cloud Platform, which follows ISO 27001, 27017, 27018 and 27701, SOC 2 and SOC 3, and GDPR-aligned data protection practices. Those are Google's standards for the infrastructure. Semantyr's own audits and certifications will be listed here when they exist.
Which language model do you use, and does that provider keep our data?+
We mainly use Google's Gemini models. The language model sees the measured results at the final step, to put them into plain sentences. It is configured so that your content is not retained by the model provider. Customers on a Custom subscription can use a private model of their own.
Is Semantyr a wrapper around ChatGPT?+
No. A mathematical model does the grouping, counting and measuring. A language model, the kind of AI behind ChatGPT, only writes up the results. There is no open prompting.
Can we run it privately or offline?+
The architecture can support private and offline deployments, depending on your needs. Tell us your requirements and we will say honestly what is possible today.
Can you meet our procurement requirements?+
Send them to us. We will tell you what we can offer today, and what is on the way.

Have a compliance question?

Reliable, repeatable, human-controlled analysis starts with knowing where your data goes. Ask us anything.

Talk to us
Read the Privacy Policy →