Three plans. All include the full detector pack, OCSF v1.3 emission, SIEM connectors, and the 30+ policy template library. Volume scales with the number of registered AI agents — not with your usage of those agents.
For developers and small teams getting started with AI agents.
For production agent fleets up to mid-size.
For regulated industries, sovereign-cloud, and 100+ agent fleets.
An agent is one logical AI service that you register with SkyDaemon — typically a (model, system-prompt, capability-set) tuple deployed to a specific environment. A LangChain orchestrator counts as one agent. A multi-agent CrewAI flow counts as one agent per registered crew. We do not bill per LLM call, per token, or per finding.
No. The Team and Enterprise plans bill solely on registered-agent count. Findings, invocations, tokens, SIEM events, and storage are unmetered.
Yes — Team and Enterprise both include a 30-day trial against your own production agents. We onboard you, configure the SIEM connector, and walk through the first findings.
SaaS is in eu-central-1 (live) with us-east-1 planned. Enterprise plans get region pinning, BYOK, and self-hosted distribution (Wave 49). FedRAMP Moderate boundary plan targets 2026-Q4. AWS GovCloud / Azure Government / GCP Assured Workloads supported through self-hosted.
SOC 2 Type II report and ISO 27001 / 42001 certifications are in progress. Available under NDA for prospects. Standard EU SCCs and DPA shipping with every Enterprise contract.
SkyDaemon imports policy DSLs from NeMo Guardrails (Colang) and red-team corpora from PyRIT, Garak, Promptfoo, Inspect AI. SIEM continuity is automatic via OCSF v1.3 — your existing Splunk / Sentinel / Falcon dashboards continue to work. Migration playbooks for each cohort vendor available on request.
The SDKs (skydaemon-py, skydaemon-ts, skydaemon-go, skydaemon-java, skydaemon-dotnet, etc.) are MIT/Apache-2.0 dual-licensed. The eval-harness is Apache-2.0. The detector pack and control plane are commercial.