Attack & integrity
Prompt injection, model manipulation, guardrail changes, and integrity failures.
Briard-AI security add-on
Know when AI behavior becomes an incident.
KAIDAN is real-time detection and incident response for AI and LLM environments. It turns approved metadata into categorized, weighted P1–P4 alerts, then carries the responder into an evidence-cited investigation without silently becoming an inline enforcement control.
Try the fictional incident
11 minutes, 47 seconds · Narrated walkthrough with captions · No sign-in needed
Free Aug. 13, 2026–Aug. 13, 2027 UTC Metadata-only by default Real-time P1–P4 alerts Tenant scoped
What KAIDAN is
Detection, response, and forensics for the AI layer.
Briard-AI and KAIDAN are products of Unfettered Minds LLC. Briard-AI answers how an AI system is governed. KAIDAN tells the response team what needs attention and what the available evidence says happened. Briard-AI provides the customer, commercial account, identity, and add-on authorization experience; KAIDAN remains authoritative for alerts, situations, evidence, cases, findings, connector receipts, and custody verification.
Live alerting
The responder sees what matters first.
KAIDAN weighs severity, confidence, asset importance, spread, and business impact. Every score stays explainable, and every alert can be acknowledged, correlated, promoted to a case, or resolved by an authorized person.
Access & data
Unexpected identities, sensitive-data access, retrieval exposure, and permission changes.
Agent actions
Tool use, autonomous steps, unsafe execution paths, and actions outside an approved boundary.
Misuse & operational impact
Abuse, anomalous volume, service impact, cost spikes, and harmful business outcomes.
Launch-year price protection
Free now, with a published ceiling for the following year.
KAIDAN has no platform fee during the fixed launch-year window. If your organization separately elects to continue after August 13, 2027, the KAIDAN platform fee will not exceed $399 USD per organization per month through August 12, 2028. Access does not automatically convert and Briard-AI will not create a charge without a separate offer and acceptance. Customer-selected infrastructure, source products, and optional professional services are outside this platform-fee ceiling.
What it does
Move from a live signal to a defensible response.
Detect and prioritize
Turn approved metadata into explainable P1–P4 alerts, grouped into four AI-specific incident categories instead of an undifferentiated event stream.
Separate evidence from inference
Keep observed facts, source assertions, detector inferences, reviewer conclusions, gaps, and contradictions visibly distinct.
Preserve verifiable custody
Retain tenant-scoped, tamper-evident evidence and produce export packages whose integrity can be checked independently.
How it works
Alert, decide, investigate, and preserve.
- 01
Observe approved metadata
Connected sources send metadata with attribution, timestamps, and integrity context.
- 02
Prioritize the alert
Explainable scoring assigns P1–P4 priority and one of four AI incident categories.
- 03
Correlate and respond
Related alerts become a situation; responders acknowledge, hand off, investigate, or promote to a case.
- 04
Preserve and verify
Teams create custody-verifiable packages for authorized review and offline integrity checking.
Published source catalog
Sixteen adapter packages, with evidence maturity kept visible.
KAIDAN ships source-specific and standards-based adapter packages. A listed package means the integration contract and tests exist; it does not claim vendor certification or a customer connection that has not produced a receipt.
AWS AI audit · Azure AI audit · Google AI audit · Google Model Armor · Prisma AIRS
OpenTelemetry GenAI · Datadog LLM Observability · OCSF over HTTPS · Kafka OCSF
Pinecone · Qdrant · Weaviate · Milvus · pgvector · Elasticsearch · OpenSearch
Connection state remains explicit: configured, tested with a receipt, degraded, or not connected. Compatible-but-unverified sources remain labeled as such.
Review source setup and validationWho it is for
Teams responsible when AI and LLM systems become incidents.
Security and incident response
Reconstruct AI and LLM activity without treating an alert as proof.
AI platform and engineering
Give responders source context, connector health, and ownership without copying application code.
Risk, legal, and compliance
Review evidence-cited cases and acknowledged unknowns while specialists retain decision authority.
Honest boundaries
Evidence first. Claims stay disciplined.
- Normal intake is metadata-only; raw prompts, responses, files, and secrets do not belong there.
- Restricted forensic capture is separate, disabled by default, and requires its own controls and approvals.
- KAIDAN observes and preserves evidence. It does not silently block AI systems or make response decisions.
- Missing source receipts remain unknown; an alert or detector result is not presented as established fact.
- Customer deployment, provider credentials, external notarization, and independent review are separate readiness states.