GovernorAIby SentinelLayer Request a briefing
Find and qualify
Discovery & Shadow AIInventory the AI activity nobody registered AssuranceEvaluate an agent snapshot before production
Decide and enforce
Policy & GovernanceKeep the policy engine you already run Runtime EnforcementDecide at the action boundary Inference & Model GovernanceGovern the model call itself Multi-cloudOne policy, each cloud's own controls
Prove and observe
Evidence & ComplianceTamper-evident record, mapped to controls Decision ObservabilitySee the decision behind the action
Risk and security
RiskScore the action, and act on the score ContainmentStop a runaway run before it lands Security architectureHow the control plane itself is built
By persona
SecurityPer-call authorization and stop controls GRCEvidence chains and control mapping Platform & AI EngineeringGovern without touching the agent
By risk
Shadow AIUnmanaged agents and connections Data leakageIn prompts, completions and tool calls Destructive actionsChanges to systems of record
By platform
AWS BedrockInbound path, guardrail attachment ServiceNowNow Assist discovery and scoped control SalesforceAgentforce and Einstein Bots, in-org MCPRegistry, proxy, argument-aware inspection
By cloud and runtime
AzureConnect, discover, govern at the gateway Google CloudValidated connect mode, bootstrap session DatabricksCluster-side egress, pinned at create Kubernetes & sidecarIn your cluster, on the agent host

DESIGN PARTNER PROGRAMME

Six weeks, one workflow, named success criteria.

We are looking for a small number of organisations running agents that do consequential things — move money, touch clinical or customer records, change production infrastructure — and who would rather govern that now than after an incident makes the case for them.

Scope

One workflow, not the estate

We pick a single agent workflow that matters and govern it properly, rather than instrumenting everything shallowly. If it does not work on one, it will not work on forty.

Who

Founders in the room

Not a success manager relaying questions. The people who wrote the enforcement path are in the working sessions, and they can change it.

Bar

Named success criteria

Agreed in writing at the start, so both sides can tell at the end whether it worked. A partnership that cannot fail cannot tell you anything.

THE TERMS

The licence is waived. The infrastructure is not free.

Stated plainly, because a partner who discovers the real shape of this in week three is a partner we have already lost.

The GovernorAI licence
Waived for the engagement. You do not pay us to run the pilot.
Infrastructure
Real, and yours. Whatever the deployment consumes in your own cloud account — compute, storage, the evidence trail — you pay your provider for, at your rates. We do not mark it up and we do not absorb it. It is usually modest for one workflow, and we will size it with you before you commit rather than after.
Production
Priced separately, and only if you decide to continue. Nothing about the pilot obliges you to, and nothing in it is priced to make stopping awkward.
What we ask in return
If it works, that you let us say so — a named case study and permission to use your logo. Contingent on the result, agreed only once the engagement has concluded and you are willing to stand behind it. If it does not work, we ask for nothing and you keep the inventory.

There is no per-seat fee, no minimum commitment and no figure to negotiate here, because the only money in this arrangement is money you pay your own cloud provider.

ADOPTION

Start read-only. Expand control where consequence is highest.

Sequenced by risk, not by upsell. Nothing enforces until you decide where enforcement belongs — and a policy can run in shadow against production traffic before it is allowed to block anything.

STAGE 1

Discover unknown AI activity

A low-friction, read-only inventory for security teams. Nothing is intercepted and nothing changes how your agents run.

Read-only
STAGE 2

Establish evidence

Align a governed action trail to what security and compliance review actually asks for, before any enforcement decision is made.

Evidence first
STAGE 3

Enforce at the action boundary

Add policy, approvals, data controls and stop controls where the integration supports them — starting with the workflow that carries the most consequence.

Where supported
Stage one, concretely Read-only discovery needs no enforcement decision and changes nothing about how your agents run.

Start with read-only discovery →See what discovery finds →

HOW IT RUNS

Four phases, and what has to be true at the end of each.

Six weeks end to end, with the targeted controls enforcing by day 30 — subject to integration readiness and your own approval to switch enforcement on.

Phase 1 · week 1
Connect a cloud account read-only and see what is actually running. Discovery is built to surface the agents, tools and MCP connections nobody registered. Whatever it finds, that inventory is yours whether or not the partnership continues.
Phase 2 · weeks 2–3
Write the policy for the chosen workflow and run it in observe mode. Nothing is blocked yet. We are establishing what normal looks like and what the rule would have caught.
Phase 3 · weeks 3–4
Turn enforcement on for that workflow. Denials, approvals held for a human, and the evidence record behind each decision. This is the week that tells you whether governance is survivable in your environment.
Phase 4 · weeks 5–6
Measure against the criteria agreed in phase one, and decide. Continue, stop, or continue with a different scope — all three are acceptable outcomes and we will say which one we think it is.

WHO THIS IS FOR

Agents that can reach data or act in production.

Healthcare and PHI, money-moving fintech, regulated decisioning, and B2B platforms with agent or MCP sprawl. If your agents only draft text, this is not urgent for you yet, and we will say so rather than take the meeting.

Apply to the programme Goes to the briefing form. We reply within 48 hours.