> ## Documentation Index
> Fetch the complete documentation index at: https://docs.getimpala.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Security & data handling

> Impala runs single-tenant, in our cloud or in yours. Your data stays isolated and isn't stored.

Every Impala deployment is **single-tenant** and logically isolated from other customers, so your data and operations remain separate and secure. That is true in both [deployment shapes](/deployment-options): an isolated deployment in Impala's cloud, or Impala's data plane inside your own AWS account and VPC, where your prompts, inputs and outputs never leave your account.

Either way, **prompts are not persisted** and **your data is never used for training** — not for model training, not for fine-tuning, not for model improvement. What Impala keeps is the shape of the workload, not what's in it: prompt lengths, concurrency, burstiness, cache reuse. That's what [Herd](/impala-herd) profiles to tune the execution plan.

Impala is certified under **SOC 2 Type II** and security-tested on a regular cadence, including vulnerability assessments and penetration testing.

<Card title="Impala Trust Center" icon="shield-check" href="https://trust.getimpala.ai">
  SOC 2 Type II status, security controls, subprocessors and documents. Self-serve and current.
</Card>

For anything specific to your environment, refer to the Trust Center or talk to your Impala contact.
