Trust and control

Automation where it is safe. Human judgment where it matters.

MIRAI does not ask companies to hand control to software on day one. Routine work is introduced gradually, important actions remain governed, and every decision can be reviewed afterwards.

The operating model, stated plainly.

Before anything is automated, we agree in writing which category each step belongs to. This table is part of the Blueprint, not a marketing summary.

Type of workMIRAI roleEmployee role
Routine and low risk MIRAI roleCompletes approved steps Employee roleMonitors
Requires judgment MIRAI rolePrepares and recommends Employee roleDecides
Sensitive or high impact MIRAI roleRetrieves supporting information Employee roleApproves and acts
Unusual case MIRAI roleIdentifies and escalates Employee roleResolves
Security principles

Least access, clear boundaries, no surprises.

Access control

MIRAI connects with the narrowest permissions the workflow needs, under your identity provider and your own approval.

Separation of duties

The system that prepares a decision is never the one that approves it. Approvers are named people, not roles shared by a team.

Change control

Workflow changes are versioned and released deliberately. Nothing about how your work runs changes silently.

Data handling

Your systems remain the record. MIRAI holds the operational state it needs to run the workflow, and only for as long as agreed.

Confidentiality

Customer data is never used to improve a shared model or exposed to another client's workflows.

Deployment options

Hosted by MIRAI in Singapore, or in your own cloud tenancy where policy or regulation requires it.

We publish certifications when we hold them, and not before. Ask us directly for the current security documentation, sub-processor list and data-processing terms.

Audit and accountability

Every action attributable, months later.

Each case carries its own history: what the system did, what it prepared, who decided, when, and on what information. This is what makes automation defensible to an auditor, a regulator or a parent.

System actions and human decisions are recorded the same way.
Approvals capture the person, the time and the reason.
History is exportable, so it survives outside MIRAI.
Case history · CASE-4818 Exportable
09:41MIRAI retrieved company records from CRM
09:41MIRAI ran credit check · result: pass
09:42MIRAI flagged 2 exceptions to policy
09:42Routed to S. Lim · finance approval required
09:44S. Lim approved 30-day terms · reason recorded
09:44MIRAI updated CRM and accounting

Implementation safeguards

01Nothing is automated before it has been observed running correctly on your real cases.
02Every automated step has a defined failure behaviour — hold and escalate, never guess.
03Any step can be returned to manual immediately, by your team, without our involvement.
04Volume limits and value thresholds are set by you, and enforced by the system.
The principle we do not bend

Automate the routine. Escalate the exception. Keep accountability human.

If a workflow cannot be run this way, we will say so before you spend anything on it.

Bring your security questions to the first conversation.

We would rather answer them at the start than discover a constraint halfway through.

Map one critical workflow