The IntegAI method
Configure the work before you configure the AI.
A governed AI use case begins with the organisational job and ends with evidence a person can review. The model sits inside that operating design.
Define the job.
Name the team, users, task and intended outcome. Decide what success can be measured during a controlled proof.
Evidence produced: Use-case statement, named owner and acceptance criteria.
Approve the sources.
Identify the policies, procedures, standards and records the Advisor may use. Exclude material that is unapproved, unnecessary or outside the proof.
Evidence produced: Approved source register and version record.
Set the limits.
Define what the Advisor may answer, what it must qualify, what it must refuse and when it must direct the user to a person.
Evidence produced: Operating boundaries, prohibited request shapes and escalation rules.
Govern the request.
Apply the configured controls and approved knowledge access before a response is released.
Evidence produced: Request identifier and governance outcome as configured.
Answer, qualify, refuse or escalate.
Release the permitted response, expose missing evidence or uncertainty, refuse the request or direct it to the accountable role.
Evidence produced: Response status, source basis and escalation direction as configured.
Preserve the decision evidence.
Record the request, source basis, governance decision, timestamps and identifiers required for review.
Evidence produced: A reviewable evidence record shaped by the buyer's use case and deployment.
The person does not disappear from the process.
IntegAI helps define where a person must remain responsible. It does not convert organisational judgement into automatic authority.
The system can make the boundary visible. The buyer remains responsible for the policy, decision, deployment and lawful use.