Security and governance for teams and agents.
Every agent and workflow runs as its own identity, with the permissions you set, and every action is logged. No static keys.
Free forever plan, no commitment, in your own cloud account.
Built from the real interface. Open the demo tenant in the live demo. No sign-in.
- Today
- 94 daysMedian time to fix a secret that leaked into a GitHub repository.Verizon, Data Breach Investigations Report, 2025A secret leaked into a GitHub repository takes a median of 94 days to fix.
median time to remediate leaked secrets discovered in a GitHub repository was 94 days
22,052 security incidents and 12,195 confirmed breaches - With Wayfinder
Wayfinder timings are from our reference deployment and demo flows, not from customer studies.
- 0 stored keysWorkflows and agents get a short-lived identity for each run. There is no password or API key to leak.
Right-size api from 6 to 3 replicas. Saves 34% on the flagged service. Needs 2 of 2 approvals from security-team.
- agent/cost-triageread_metrics, list_resources, as sa:cost-triage, in its own sandboxallowed
- agent/cost-triageopen_pull_request, gated on approvalheld
- jon@acmeapproved plan, 03:41recorded
Before and after
Before
- Long-lived cloud keys in CI and on laptops
- Access managed in a spreadsheet
- Agents with more permission than the person who wrote them
- Audit assembled from five tools before a review
- Autonomy blocked because nobody can vouch for it
With Wayfinder
- Workload identity via OIDC, nothing stored either side
- RBAC by scope: tenant, workspace, environment
- A run can never exceed its author, checked at authoring time
- One audit trail, every action attributable, agent ones included
- Own kernel sandboxes and approval gates, so security can say yes
How it works
Identity
Users through your identity provider. Workloads, workflows and agents as service accounts with scoped roles.
Limits
Tools are the permissions for agents. Policy checks before the cloud. Approval gates in the graph.
Evidence
Every run, delivery, call and approval is a record you can open again.
In the product
- WFWorkflow cost-control, run 88Input: CostFinding, api, +34% week on week. Identity: sa:cost-triage. Sandbox: own kernel.
- AICalled read_metrics(api, 7d)CPU 11% average across 6 replicas. Requests flat. Replica count changed on 14 Sep.
- AICalled list_resources(stack: payments)api: 6 replicas, requested by deploy #2231. Previous: 3.
- AIResultCause: replica bump left in place after load test. Proposed: 6 to 3, saves 34%. Tested in a replica, p95 unchanged.structured result2 tools calledheld for approval
What this shows
- The agent ran as sa:cost-triage, in its own sandbox, with two read-only tools.
- The write was held for approval; the agent could not perform it itself.
- The conversation is the audit log: every call and every result.
Components
OIDC trust between your cloud account and Wayfinder. No keys.
Roles per tenant, workspace and environment. Roles apply downward: tenant, then workspace, then environment.
The identity a workflow task or agent acts as.
An agent with a shell gets its own kernel, default-deny network, no metadata access.
Run before anything reaches the cloud.
Invocations and conversations kept as resources, attributable and replayable.
Security and governance
Accredited: ISO 27001, ISO 9001, ISO 14001, and cloud partner status with AWS, Microsoft and Google.
Runs in your own cloud account; your data does not leave it.
Every credential is scoped, purposed and rotated as an external identity.
The audit trail is the same record the platform runs on, not a copy.
Accreditations and partnerships
Book a demo for your security team.
Book a demo and we will walk through the controls with them.