Inspiration
As AI agents become autonomous in healthcare, finance, and legal systems, we realized a critical gap: there is no real-time "race control" to stop unsafe AI actions before damage occurs. Existing tools detect violations after the fact.
Inspired by the Williams Racing theme, we built CIRIS to bring real-time safety, control, and accountability to enterprise AI systems.
What it does
CIRIS is a real-time AI governance firewall that evaluates every AI agent action before execution.
Each action is classified as:
- 🚨 VETO → Blocked immediately and escalated to Jira with risk analysis and remediation steps
- ✅ PASS → Approved and logged into Confluence as an immutable audit record
This ensures privacy protection, ethical behavior, security enforcement, and compliance by design.
How we built it
CIRIS is built natively on Atlassian Forge and integrates directly with:
- Jira for automated governance review tickets
- Confluence for tamper-proof audit logs
- Rovo to explain why actions were blocked or approved
External AI agents communicate with CIRIS via secure web triggers. A decision engine evaluates risk in real time and enforces governance outcomes within milliseconds.
Challenges we ran into
- Designing a governance layer that operates before AI actions execute
- Creating meaningful, human-readable risk explanations
- Balancing speed (<500ms) with deep safety checks
- Building a system flexible enough for multiple regulated industries
What we learned
We learned that AI governance must be proactive, not reactive. Integrating governance directly into operational workflows (Jira + Confluence) dramatically improves trust, adoption, and accountability.
What's next for CIRIS
- Industry-specific policy packs (HIPAA, GDPR, SOX)
- Agent behavior trend analytics
- Automated compliance reporting
- Expanded Rovo-driven investigations
CIRIS enables organizations to deploy AI at speed — without losing control.
Log in or sign up for Devpost to join the conversation.