GuidesBy roleFor leadership

For leadership

What you are buying

Not an AI tool. A record of how your organization actually operates, and a gate in front of everything the AI in it does.

The gate is the part that lets you say yes to AI adoption without the answer to "what stops it doing something stupid?" being a hope.

The weekly read, four numbers

Console → Dashboard, the summary view an executive sees on sign-in

Console → Dashboard, the summary view an executive sees on sign-in

screenshot slot, save as public/screens/…png and set  src

Four numbers, and each links to the screen that explains it. Anything not on this list is somebody else's daily view.

1. Open incidents, by severity

Should be near zero and trending down. A critical incident sitting open for days is a staffing answer, not a tooling one.

2. Approval latency

How long a request waits for a human. This is the leading indicator. Rising latency means governance is routed to people who are not looking, and the ending is always the same: somebody removes the rule.

3. Governed actions

How much AI activity is actually running behind the gate. A number that is not growing while your engineering team says it is adopting AI means the gate is not where the work is.

4. Decisions approved

How much of the organization's tacit knowledge has become explicit. A queue that nobody works means the understanding is not being built.

Questions this answers well

  • What did we decide about X, and who decided it?
  • What has AI actually been doing in our systems this month?
  • Which teams' work depends on which others'?
  • If this person left tomorrow, what only they know?
  • Can we prove to a customer that an AI cannot touch their data unsupervised?

Questions it deliberately refuses

  • Who is my most productive engineer? The data is right there and would be wrong: the graph sees the tools that are connected, not the work that happened elsewhere. Memnox is an operational record, not a performance management system.
  • What will happen next quarter? Memnox reports what happened. A forecast is a model's opinion, and there is no model in this path.
  • Is this code any good? Out of scope, permanently. Memnox governs agents; it is not one, and it never reviews code.

What adoption actually looks like

Phase

Weeks 1 to 2

Everything observed, nothing enforced. You are reading

Weeks 3 to 6

One or two rules enforced. Review queue worked weekly

Months 2 to 3

Approvals routed and answered. Incidents have owners

Month 4+

Default-deny on ungoverned actions. Audit export in the compliance pack

If a rollout stalls, it is nearly always at the same place: nobody owns the review queue. That is one named person, fifteen minutes a week.

The two questions worth asking your team

"Show me the last thing Memnox stopped." If the answer is nothing, either nothing risky is happening or the gate is not in the path. Both are worth knowing.

"Show me an approval from last week." It should name a person, an action, a target and an environment. If approvals are being granted in bulk by a shared account, the record is worth nothing and that is fixable today.

Cost shape

The hosted control plane is per-seat with metered usage on top: governed actions, extraction runs and meeting bots. Current plans and prices are on the pricing page, and what your organization is actually consuming is on Settings → Billing.

The runtime is open source and free, forever, on your own machines. An organization can govern every agent it runs without paying for anything; the cloud is what adds organizational understanding, review and reporting on top.