
Access provisioning and approval routing
Access requests arrive by chat, get granted by whoever is around, and nobody remembers who approved what. An agent records each request, routes it to the system owner, and passes approved ones to IT. Approving and granting stay with people.
The short answer
An agent routes access requests by recording each one with the system, the level of access and the reason, checking it against the requester's role, and sending it to the named owner of that system for approval.
Approved requests go to IT to grant. The agent grants nothing, and a second approver signs off on anything privileged.
Why does nobody know who approved that access?
Someone needs access to the finance tool, so they ask in chat. Whoever has admin rights grants it, because asking the owner would take a day. Six months later the person has moved teams, still has access, and nobody can find the message where it was approved, if it was. Multiply that across every tool a company adds, and the access list stops matching anyone's idea of who should have what.
Who is this for?
An IT lead, systems administrator or operations manager at a company with a growing set of tools and no identity team. It suits businesses that can name an owner for each system and describe, at least roughly, what access each role needs. Granting the access and deciding exceptions stay with people.
What changes when every request has a path?
| Part of the request | Granted on request | Routed by an agent |
|---|---|---|
| Where it is asked | Chat, email, in the corridor | One request record, whatever the channel |
| Who approves | Whoever has admin rights | The named owner of that system |
| Fit with the role | Not checked | Compared with the role's usual access, with any difference shown |
| Privileged access | Treated like any other request | A second approver required |
| The record | Lost in a chat thread | Who asked, who approved, who granted, when |
How it works, step by step
The agent carries the request from asking to approval. Approval and granting are done by people, in that order.
Take the request in any channel you name
Check it against the role
Route to the system owner
Add a second approver for privileged access
Human gate: owners approve and IT grants
Track temporary access
Prepare the periodic access review
What does the agent never do with access?
Grant or remove access
IT does, after approval.
Approve its own suggestions. It shows the role comparison and leaves the call to the owner.
Treat silence as yes
Unanswered requests are chased, never assumed approved.
Hold admin rights
It needs read access to request records and role definitions, nothing more.
Override an owner. A refused request stays refused unless the owner changes their mind.
Where this fits
From request to granted access is a single process, and Tier 1, Automate a Workflow covers it. Your system owners and role definitions are what the readiness checklist will test first. The routing pattern is the same one used in purchase approval workflows that don't stall, and requests that are not about access go through internal request routing and triage. The one-name rule for each approval comes from approval gates. The use case library has more.
How does access routing differ from general IT tickets?
| Question | General IT ticket | Access request |
|---|---|---|
| Who decides | IT, usually | The owner of the system |
| What it changes | A device or a setting | Who can see and change information |
| What must be kept | The fix | The approval, for later review |
| When it ends | When fixed | When access is removed |
Related reading
- Purchase approval workflows that don't stall: the same request-to-approver routing for spending.
- Internal request routing and triage: where other internal requests go.
- Exception detection and routing across your ops stack: spotting access that does not match its approval.
