Trust center
A security model built for regulated environments
Grafty is single-tenant and customer-operated. We deliver software and licenses; your team runs the platform inside your own network boundary. We do not host your apps, your source, or your data.
Data boundary
Where data lives
Inside your network
Builder UI, source code, Git history, previews, app data, audit logs, integration credentials.
Configurable egress
LLM provider calls (OpenAI, Anthropic, Google) or a local in-network model endpoint — chosen per project.
Never on our servers
We do not receive, store, or process your prompts, code, previews, or runtime data.
Single-tenant, customer-operated
Each customer runs their own Grafty instance on their Linux infrastructure. There is no shared SaaS plane — nothing about your deployment is co-located with another tenant.
A Kubernetes namespace per app
Every generated app runs in its own namespace with resource quotas (size tiers XS–L). Deletion cleans up every resource — no orphans, isolation by default.
Access control
RBAC and immutable audit
Roles
instance_admin, developer, and viewer. Membership and limits are scoped by team.
Audit
Append-only audit events for privileged actions, scoped by role and exportable for SIEM ingestion.
Governance
LLM spend controls
Token-usage dashboards by user, project, and team. Per-user, per-project, and per-team budgets. Hard stops block generation before another provider API call is made — preventing runaway invoices. You bring your own keys, and Grafty never marks up usage.
Encryption
Data in transit and at rest
TLS everywhere
All inter-service and ingress traffic runs over TLS.
Secrets at rest
Provider keys and integration tokens are encrypted on disk on the instance.
Short-lived sessions
Signed session tokens with rotation.
Subprocessors
Customer-controlled, not Grafty
The external services a deployment can reach are chosen and contracted by the customer, using the customer’s own accounts and keys.
| Service | Purpose | Controlled by |
|---|---|---|
| OpenAI / Anthropic / Google | Optional cloud LLM inference | Customer (your keys) |
| Local / on-prem LLM | Optional inference inside your network | Customer |
| Jira, Slack, GitHub, GitLab, Notion | Optional connectors | Customer (your tokens) |
Security review packages
Architecture overview, threat-model summary, sample questionnaire responses, SBOM, and a reference DPA are available on request. Contact us to receive the latest package.
Vulnerability reporting
Email info@grafty.ai with the subject “[Security]”. We respond within one business day.