Resources / Professionals & Organizations / Explore
Applied System Patterns: Intake and Routing
Generalized patterns for moving a request from first contact to the right person, with the failure modes each one carries.
Intake is where most downstream operational problems are created. These are generalized patterns — no client-specific detail — with the conditions each suits and the way each tends to fail.
Single queue, human triage
Everything arrives in one place and a designated person routes it. Simple, transparent, and easy to change.
Fails when volume exceeds one person's attention, when the triager becomes a single point of failure, or when the routing logic lives only in that person's head.
Structured intake with rule-based routing
The requester answers a defined set of questions and defined rules route the result. Consistent and auditable.
Fails when the questions are written for the organization's convenience rather than the requester's understanding, and when nothing catches the cases the rules did not anticipate.
Tiered routing with an exception lane
Common cases route automatically; anything ambiguous, high-acuity, or rule-breaking goes to a named human lane.
This is usually the more durable pattern. It fails when the exception lane has no capacity, no timeframe, and no owner — at which point it becomes a place requests go to wait.
What to instrument regardless of pattern
- Time from first contact to first human response, and to the first substantive action.
- Rate of re-routing after initial assignment — a direct measure of routing accuracy.
- Volume and resolution time in the exception lane.
- Requests that entered and never reached a disposition.
Where AI is sometimes proposed here
Classification and summarization of inbound requests are common proposals. They can reduce reading time, but they sit directly upstream of a consequential decision about who a person sees and how quickly.
If used at all, the pattern that holds is assist-only: the model organizes and proposes, a person confirms the routing, high-acuity and exception paths are never auto-dispositioned, and disagreements between the proposal and the human decision are logged and reviewed.
Continue
Documentation Governance: A Working Checklist
Questions to work through when documentation quality is inconsistent and no one is sure where the system breaks.
Read5 min working sessionHuman-Led AI Governance for Behavioral Health & Human Services
An interactive walkthrough of a governed AI workflow, from current state to a practical next step.
Open the experience8 min interactiveNext step
Working through this in a specific organization?
A defined solution is not required. A focused consultation can start from the current context and existing strengths, and shape a practical next step.