Helix Foundry for CISOs

Helix Foundry can run entirely in your infrastructure, on your Kubernetes in your cloud account or on-prem, with your own models or none of ours. Connectors are read-only, credentials are sealed when entered, nothing is visible until it is granted, and agents act with their user's role and need confirmation for every change to your setup.

Book a meeting

Where data lives

  • Your infrastructure. Deploy on your own Kubernetes, in your cloud account or on-prem. Your data stays inside your network.
  • Helix Cloud. Managed by Helix, with workspaces and member roles.
  • One computer. Self-hosted locally, Foundry listens on 127.0.0.1 only, and data stays on the machine unless you pick a hosted model.

Which AI sees your data

  • Your models and keys. Self-hosted, use Amazon Bedrock, Azure OpenAI, Claude, OpenAI or any OpenAI-compatible server on your own accounts, or self-hosted models such as Ollama.
  • Or none of ours. Your own agents query the graph over MCP in SQL, Cypher or JSON.
  • On Helix Cloud. Helix runs Anthropic's Claude models for you, or your own Anthropic key on Pro and Max.

Access control

  • Nothing visible by default. A member, team or agent key sees nothing until it's granted a source.
  • Hidden in the graph. Entities, facts and rules built from a source a person or key can't read are hidden from them.
  • Directory sync. People and groups sync from Microsoft Entra ID or Okta.

Credentials, connectors and agents

  • Sealed credentials. Passwords and API keys are sealed when you enter them. Agents never see them.
  • Read-only connectors. Connectors read your systems and never write back. Later syncs read only what changed.
  • Agents act as their user. Through the Helix MCP, an agent has its user's role, and every change to your setup needs that user's confirmation.
  • Traceable answers. Every answer shows each step the agent took and cites the records, rows and quotes it used.

Assurance

  • SOC 2 trust center. Compliance details are at trust.helix-db.com.
  • Open source. Apache 2.0, so your team can review the code it deploys.
  • Security review support. Enterprise plans include security review support and a named contact.

FAQ

Does our data leave our network?

Not when Foundry runs in your infrastructure with self-hosted models, or with no Foundry model at all. A hosted model you choose, such as Amazon Bedrock or Azure OpenAI on your own account, receives what Foundry sends it.

Does Foundry write to our systems?

No. Connectors read your systems and never write back, and later syncs read only what changed.

How are AI agents controlled?

An agent key sees nothing until it is granted a source. An agent connected through the Helix MCP has its user's role, so it can only do what that user could do, and every change to your setup needs that user's confirmation. Agents never see the credentials Foundry stores.

Is the free local install safe for a team?

No. It is for one person on their own computer and has no sign-in, so never expose it through a proxy, tunnel or port forward. For a team, use Helix Cloud or deploy Foundry in your infrastructure.

How do we start a security review?

Book a meeting at https://cal.com/team/helixdb/quick-chat. Compliance details are at https://trust.helix-db.com, and the source is at https://github.com/HelixDB/helix-foundry, so your team can review the code before a proof of concept.