Sattva builds custom knowledge systems for nonprofits. We are a builder and maintainer, not a hosting vendor. This page explains what that means in practice and what it means for your security and data governance reviews.
The short version
- You own your system. The system we build is yours: your data, your configuration, and any code written specifically for you. It runs under your control and continues to do so should you choose to stop working with Sattva. The shared components underneath it are Sattva’s, licensed to you permanently (see point 4).
- It runs in your environment. Your system is deployed to accounts you control (your cloud, your data warehouse, your LLM provider). Sattva does not host, store, or process your data on Sattva infrastructure.
- We access it as your contractor. This is the forward-deployed engineering model: our engineers work inside your environment, on your systems, rather than pulling your data into ours. We use credentials you issue, under your access policies, with the same logging and controls that apply to your staff. You can revoke that access at any time.
- We keep the shared components. Sattva maintains a common library (integrations, schemas, workflows) used across every system we build. Those components are code, not data, and nothing in them is derived from your data. You receive a perpetual, royalty-free license to every component in your system. Our agreement with you names which categories count as shared components; everything outside them is yours.
- You can leave at any time. The codebase lives in a repository you own, and you can modify or extend it without us. If you keep working with Sattva, we maintain your system under a maintenance agreement, which at minimum keeps the shared components current and can extend to building new custom capabilities.
What this means for your review
Because your data never leaves your environment, Sattva is not a data processor or a platform vendor in the usual sense. The relevant security certifications (SOC 2, encryption, access logging) are those of the platforms you already run on. Our obligations are those of any contractor with system access: we follow your policies, we appear on your authorized-user list, and we are bound by the confidentiality terms in our agreement.
Concretely, in a vendor review we expect to be classified as professional services with system access, not as a hosted software vendor. We are happy to sign your standard contractor confidentiality and acceptable-use terms, and to complete a security questionnaire scoped to that role.
What we don’t do
- We don’t host your data, or any copy of it, on our own systems.
- We don’t pool, analyze, or learn from client data across clients. What we reuse across clients is code and design patterns, never data.
- We don’t run monitoring that sends your data to us. Any observability we build stays inside your environment.
- We don’t retain access after an engagement ends unless you ask us to under a maintenance agreement.
Where the line is
If you ask us to run something for you on our side (a hosted service, a shared model endpoint, a central dashboard), that changes our role, and we would treat it as a separate agreement with its own data terms.
Questions
Ask us anything about this. Our goal is to align on security and data governance, and we are open to modifying our approach in order to meet your internal governance policies.
All resources →