All your data resides within our cluster within secure data centers. We back up our DBs daily. No data is shared or sent outside our cluster without the customer's consent.

Yes, Trebellar is SOC2 Type 2 certified.

Yes, all our ML models (e.g., anomaly detection, predictions, etc.) run in our own data centers without the need for external providers. We also host infrastructure that leverages open-source LLMs (e.g., Llama from Meta).

Every model that requires customer data runs and trains independently for each customer. By design, no data co-mingling ever occurs.

As a general rule, Trebellar does not require any PII information to deliver on CRE use cases. When PII is used, it's only by customer request — and we have a layer of abstraction that prevents any of that data from being used by AI/ML models.

We use Enterprise versions that forbid using the data for training or other commercial purposes. When interfacing with external LLMs, our systems precompute calculations and structure data to ensure reliable, structured responses.

The LLM will never have context or identifiers associated with your organization. We use a layer of abstraction that de-identifies and abstracts any specific data pertaining to your organization. In plain terms, the LLM has no knowledge of you or any of your company-identifying attributes.

No, our systems do not pass identifiers, such as company name, to 3rd party models. We use an abstraction layer to strip out all company-identifying metadata.

Yes, please speak with your Trebellar account manager to discuss your options.