Trust
Hire it. Audit it. Cap it. Fire it.
Enterprises don't fail with AI because models are weak — they fail because nobody can answer what an AI employee did, on whose authority, and how to stop it. This page is that answer, in the same terms we hold ourselves to internally.
The four proof points
What governance means here, expanded.
Permissioned
An AI employee can only use the permissions you granted it, for your organization, and nothing ambient.
Access is default-deny — a role is granted permissions explicitly, never assumed.
Audited
Every action an AI employee takes is written as an append-only record: who, what, when, and through which channel.
Backed by a dedicated audit event log, not application logs an operator could quietly edit.
Capped
Usage and spend enforce hard limits rather than best-effort throttling, and automation ships default-off.
Limits are checked again at approval time, not just when a job is first scheduled.
Stoppable
Any role can be paused or revoked instantly — no cascading cleanup, no waiting on a support ticket.
Kill switches and readiness checks are built into the platform, re-checked on every request.
How availability labels work
Availability is verified, not marketed.
Every product and role carries a status verified against what is actually running, shown as one of three public labels. A label changes the moment the underlying status changes — it is never a hand-set marketing claim.
Running in production today. Ready to license.
0 of 9 workforce roles today.
Production-ready, released to approved customers only.
0 of 9 workforce roles today.
Not yet available. Being built or specified — not offered for license yet.
9 of 9 workforce roles today.
Only "AVAILABLE" and "EARLY ACCESS" can be licensed; "IN DEVELOPMENT" cannot, and no product or role is ever announced past what it has verifiably earned.
Honest by default
What we disclose, and why.
An inventory that silently drops a known gap has the same credibility problem as one that overstates. Where something exists but isn't ready — or turns out not to exist at all — that is written down rather than quietly omitted.
Nothing is sold ahead of its status.
A product that is still In Development cannot be sold regardless of interest or payment; the check is enforced in the platform, re-checked at the moment of approval, not just at listing time.
In-development roles show the label, not a demo.
A role marked IN DEVELOPMENT never carries a capability walkthrough beyond what it can honestly do today — see the roster on /workforce.
Compliance
What's enforced today, not promised.
Append-only audit trail
Every audited action records the actor, the organization, and the target, and cannot be edited after the fact.
Licensing only when ready
Only products labeled Available or Early Access can be licensed, and only Available products can go into live production use. The platform itself enforces this.
Human sign-off stays human
Regulated determinations are made by licensed professionals; an AI employee drafts, a person owns the conclusion.
Pax Governance’s records — decisions, tasks, authorizations, check results, and verification records — are append-only, enforced by the database, and the loop they record was proven on our own production system in October 2026.
Read what was proven, and what was not →Platform governance module →
Organization
Who is accountable for this.
We own it
Pax Equitas holds the source, the IP, and the roadmap for every product in the catalog. Nothing here is a reseller relationship.
We operate some of it
Several products run inside our own operating businesses before they are ever offered to a customer — that is where a defect is found first.
We license the rest
Anything that survives our own use can be licensed individually, bundled, or deployed under a purchasing organization's own brand.
Leadership
Accountability, not a distributed anonymous team.
Pax Equitas is a founder-led product house: the same organization that builds each product is the one that operates it inside real operating businesses first. There is a single accountable owner for the roadmap and the governance posture described on this page — not a committee, and not an outsourced team.
Employed Under Governance
Permissioned
Every action runs under permissions you granted, never ambient access.
Audited
Every action writes an immutable audit event.
Capped
Usage and spend enforce hard limits, checked at approval time.
Stoppable
Any role can be paused or revoked instantly, no cascading cleanup.
AI Workforce roles are software — permissioned, audited, and stoppable systems, not employees, contractors, or legal persons.
AI Workforce roles are software — permissioned, audited, and stoppable systems, not employees, contractors, or legal persons.
See the governance model in your own instance.
Every claim on this page is something you can check against a running system, not a slide.