Share
AI Agent Governance

One Word. Full Governance.

Custom Commands Turn Domain Knowledge Into Governed Automation.

Your best operators workflows live in their head. When they leave, the knowledge leaves with them. IdentityRM lets customers encode domain knowledge into governed commands that anyone can run. No coding. No runbooks. One word.

Martin Gee 4 min read
custom-commands workflow-automation governance MCP business-process
Martin Gee

"A portal takes 7 steps and a specialist. A prompt takes one sentence. A governance command takes one word. Each layer eliminates not just time, but the need to know how the systems work underneath. That is governed speed."

— Martin Gee, Founder & CEO, IdentityRM

Your best operator knows how to onboard a Seattle vendor in 4 minutes. They know which systems to hit, which scope to set, which approvals to skip for standard tier.

That knowledge lives in their head. Maybe in a runbook — if you are lucky.

When they leave, the runbook becomes archaeology. The next operator reads it, guesses, gets it wrong, opens a ticket. The one after that skips the runbook entirely because nobody told them it existed.

This is the bus factor applied to governance workflows. And every enterprise has it.

What if the workflow itself was a product — governed, reusable, and available to anyone with the right persona?

The Prompt Advantage

The first step is replacing the portal with a prompt.

Instead of navigating a custom-built vendor admin screen — the one that only Dave understood, the one nobody maintains anymore — the operator just says:

“Provision Acme Corp vendor accounts for the Seattle distribution partnership.”

IdentityRM resolves intent, checks scope, delegates authority, logs the decision. Four seconds. Every step governed. Full decision trace.

150 vs 40
Words per minute: speech vs typing

A Stanford study confirmed speech is 3x faster than typing. Your prompt-based interface is not just simpler — it is physically faster to operate. Operators dictate a sentence in 3 seconds that would take 10 to type.

But the prompt still requires the operator to know the workflow — what to ask for, how to phrase it, what scope applies. A new hire stares at the prompt the same way they stared at the portal. The input is faster. The knowledge requirement is identical.

The prompt eliminated the portal. It did not eliminate the knowledge requirement.

That is what commands do.

Governance Commands

Here is the reveal: customers can create their own commands that chain governance operations into a single governed workflow.

No code. No orchestration framework. No YAML pipeline. Just a command definition that encodes what your best operator already knows:

/onboard-seattle-vendor
Customer-defined governance command
desc Standard vendor onboarding for Seattle partnerships
persona Partner Admin
step 1 resolve_partner → match vendor to partnership tier
step 2 check_scope → verify operator authorized for Seattle delegation
step 3 delegate_authority → Standard Vendor role, 90-day review
step 4 log_decision → full decision trace

That is the entire definition. A customer admin creates it in minutes. No developer involved.

Now watch it run:

"Onboard Seattle vendor" one sentence
Persona: Partner Admin / Seattle
✓ resolve_partner → Acme Corp → Standard Tier
✓ check_scope → Operator authorized for Seattle vendor delegation
✓ delegate_authority → Standard Vendor role, scope: Seattle, 90-day review
✓ log_decision → Decision trace #DT-5102 recorded

One word triggered four governed operations. All persona-scoped. Full decision trace. The operator did not need to know which systems are involved, which API to call, or what scope rules apply. The command encoded that knowledge. The platform enforced the guardrails.

The Dave Problem — Solved at a New Level

Remember Dave? The operator from Boulevard of Broken Dreams who left in 2023 and took the portal knowledge with him?

Dave’s runbook becomes a command. Dave leaves. The command stays. Anyone with the Partner Admin persona runs it. The knowledge is not in Dave’s head — it is in the platform.

But this goes further than preserving Dave’s knowledge. Commands compound. The vendor onboarding team creates /onboard-seattle-vendor. The compliance team creates /quarterly-vendor-review. The offboarding team creates /offboard-vendor-cascade. Each command encodes a different specialist’s knowledge. Together, they form a governed library of institutional workflows that survives any single departure.

The command does not bypass governance. It rides on top of it. Same persona-scoping. Same decision traces. Same audit trail. The speed comes from encoding the workflow, not from skipping the guardrails.

Think of it as a governed macro. The command encodes what to do. The platform enforces who can do it, to whom, and within what scope.

The Guardrail Inversion

Same command. Different operator. Different result.

Partner Admin / Seattle
/onboard-vendor
✓ Resolves Seattle vendors only
✓ Scope: Seattle distribution
✓ 90-day review cycle
✕ Other regions invisible
Partner Admin / Global
/onboard-vendor
✓ Resolves all vendors, all regions
✓ Scope: global distribution
✓ 180-day review cycle
✓ Cross-region visibility

Same command. The persona determines what happens.

This is the inversion. Governance is not a constraint that slows you down. It is the rails that let you go faster. Without guardrails, every command would need manual scope verification — a human checking that the operator should be doing what they are doing. With architectural guardrails, operators run commands at full speed knowing the boundaries cannot be crossed.

Runbook Orchestration Layer Governance Command
Who creates The specialist A developer Any admin — no code
Who can run Whoever finds it Systems it is wired to Anyone with the right persona
Guardrails Hope and documentation Hardcoded in the workflow Architectural — enforced per persona
When Dave leaves Knowledge walks out Code rots without maintainer Command stays. Governance stays.

Orchestration layers encode workflows. IdentityRM encodes governed workflows. The governance is not overhead. It is the product.

One Word

Portals required training. Prompts require intent. Commands require one word.

Your fastest operator on Day 1 is faster than your specialist was on Year 5 — because the specialist’s knowledge is encoded in the command, and the guardrails are architectural. The compounding effect is real: every command created is institutional knowledge preserved, every correction improves the underlying model, every new persona-scoped variant extends the library without duplicating logic.

Stop building orchestration scaffolding. Start encoding governance.


IdentityRM lets customers create governed commands — no code, no orchestration frameworks, full persona-scoped enforcement. See the platform or request a demo.


Share
All Posts