The enforcement & compliance layer · the tools you already run · the “HOW”
Control planes & MCP gateways
Broker agent-to-tool connections · enforce access policy · keep audit trails.
AI / LLM gateways
Route model calls · block PII leakage & prompt injection · cap cost and tokens.
Runtime guardrails & observability
Discover agents · evaluate outputs · monitor drift, bias & behavior in production.
Security & threat detection
Secrets management · token masking · real-time scanning of AI traffic.
Kill switch & access control
Freeze agents or teams instantly · role-based permissions and scopes.
GRC & compliance platforms
Map AI to EU AI Act, NIST & ISO 42001 · registries & audit evidence.
Works alongside the gateways, guardrails & governance platforms you already run.
WEKID doesn’t replace them — it tells them what to enforce.
Enforcement tools
answer HOW — route, gate, block, kill, log & audit agent traffic at runtime. They enforce whatever policy they are given.
TrustBridge
WEKID
answers WHAT & WHY — defines the authority, accountability & epistemic standard those policies must encode.
WEKID — the complementary layer · the “WHAT & WHY”
Defines what AI should do
The epistemic framework — Wisdom → Experience → Knowledge → Information → Data. Higher layers gate lower ones; a strong score can't rescue fabricated data.
Epistemic framework
Assigns decision authority
Five delegation levels — Human Authority, Augmented, Supervised, Constrained, No Authority — that become the policy a control plane enforces.
Decision authority
Scores epistemic maturity
Model One turns knowledge maturity into a Maturity Score, so trust is measured — not assumed — before authority is delegated.
Maturity Score
Designs accountability
Names who is accountable and on whose authority an autonomous agent acts — the human-authority model behind every enforced action.
Authority & accountability
Certifies people & partners
A vendor-neutral standard with certification & licensing — governs the professionals and organizations, not just the traffic.
Standard · certify · license
WEKID authority outputs → encoded as policy, enforced at runtime by the tools above
Human Authority
Augmented
Supervised AI
Constrained AI
No Authority
the policy every gateway is meant to enforce
How to read this: Every enforcement, risk & security tool quietly assumes two questions are already answered — how mature is the knowledge, and how much authority it should be given. WEKID is that epistemic layer beneath them.
Sources (public vendor materials, Jul 2026): Barndoor AI · TrueFoundry · Microsoft · IBM · Lasso Security · Lunar.dev · MintMCP · Arthur · Credo AI · OneTrust. Vendors named for reference only; no affiliation or endorsement implied.