What Memnox is

This page assumes nothing technical. If you work in operations, finance, legal, HR, customer success or leadership, this is the page to read, and it is the same explanation an engineer gets, not a simplified one.

The problem

A company runs on three things that do not talk to each other: people, the software they use, and now AI agents.

Slack holds the conversation. GitHub holds the change. The CRM holds the customer. Finance holds the contract. The recording of Tuesday's meeting holds the decision everybody is now arguing about. And every AI assistant somebody turned on holds its own private slice of context that disappears when the chat window closes.

Nobody, and nothing, understands the whole organization. That is what actually produces:

• work done twice, because two teams never saw each other's thread;
• decisions that contradict a decision made four months ago that nobody can find;
• approvals that were needed and never asked for;
• processes that live in one person's head until they are on holiday;
• handoffs that drop silently between two tools.

Adding another AI agent makes this worse, not better, because it adds one more fragment that sees even less than a person does.

What Memnox does

Memnox connects humans, AI agents, applications, departments, policies, goals and workflows into one continuously updated understanding of the business.

Three things follow from that, and they are the whole product:

1. It understands

Memnox reads what your organization already produces, messages, pull requests, issues, documents, meeting transcripts, and builds a picture of how the company actually works: who owns what, which customer a thread belongs to, which goal a project serves, what was decided and by whom.

Nothing enters that picture without a link back to where it came from. Every single thing Memnox believes can be clicked back to the Slack message, the pull request, or the minute of the recording it came from. See Source events and evidence.

2. It governs

Once Memnox knows the organization's rules, payments code needs a second pair of eyes, nothing touches the production database, spend over $5,000 goes to finance, it can enforce them. Not as a reminder, as a gate: an AI agent that tries to do one of those things is stopped, or held until a named human decides.

The gate contains no AI. It is a deterministic rule engine: the same request produces the same answer every time, and the answer quotes the rule that produced it. See How a decision is made.

3. It keeps work moving

With ownership, policy and history in one place, the handoffs stop being somebody's job to remember. A signed contract fans out on its own: billing is raised, engineering work is opened against the right owners, onboarding starts, marketing is briefed, leadership is notified, and the approvals policy already allows are granted without a meeting.

What Memnox refuses to do

This matters as much as the list above, because it is what makes Memnox safe to put in front of everything.

Memnox governs agents. It is not one.

• It does not write code.
• It does not review code, comment on pull requests, or suggest reviewers.
• It does not run anything in a sandbox on your behalf.
• It does not put an AI model in the path of a security decision.

It reads a pull request the way it reads a Slack message: as evidence that a decision may have been made. Forming an opinion about the code inside it is review, and review is somebody else's product.

The two halves

Memnox ships as two pieces. Most organizations end up using both, but they are genuinely independent, and it is worth knowing which one you are talking about.

The runtime is the gate

It runs on your own machines. It is open source, free, and installs in one command. It needs no account, no API key and no network connection, because deciding whether an action is allowed never required one.

An engineer installs it, and from that moment every AI action in their editor is decided before it runs.

The cloud is the understanding

It runs hosted. It is what reads your Slack, GitHub and meetings, builds the picture of how the company works, holds the queue where a human approves what was decided, and produces the reports.

An admin sets it up in a browser. Nobody needs a terminal.

How they relate

The runtime works completely on its own. A developer can govern their agents on a laptop forever and never talk to a server that is not theirs.

The cloud makes the runtime better informed: decisions your team approved become constraints the gate checks against, so a rule your organization actually agreed on starts stopping things.

Nothing about the local path changes when the rest of the company adopts the cloud. That is the whole design.