Specialist roles with boundaries, not digital people.
An Amani Asili AI employee is a bounded configuration of responsibility: it has defined tools, explicit limits, a named human owner and a clear escalation path. It prepares and coordinates; people decide.
Where records, money and leadership attention are managed.
Each role page shows what the role receives, what it prepares, what it must never do, and who supervises it.
Bookkeeper
Transaction evidence captured, classified and matched — with final posting kept behind human approval.
Inspect role boundaryPayroll Officer
Payroll inputs gathered, validated and prepared for review — pay itself is never released by the role.
Inspect role boundaryTax Assistant
Obligation calendars and evidence packs kept current, so filing periods start organised.
Inspect role boundaryExecutive Assistant
Correspondence triaged, briefs assembled and follow-through tracked — commitments stay with leaders.
Inspect role boundaryWhere employees, candidates, suppliers and data move.
HR Officer
Routine employee questions answered from approved policy; sensitive cases escalated to people.
Inspect role boundaryRecruiter
Candidates, interviews and feedback kept moving — selection decisions stay with your team.
Inspect role boundaryProcurement Officer
RFQs, comparisons and order follow-up prepared — supplier awards and spend stay human.
Inspect role boundaryOperations Analyst
Backlogs, bottlenecks and variances made visible from approved data — interpretation stays with owners.
Inspect role boundaryWhere customers, revenue and obligations are handled.
Sales Coordinator
Enquiries answered, proposals prepared and follow-ups scheduled — pricing and promises stay human.
Inspect role boundaryCustomer Care Agent
Every message triaged and routine questions answered from approved knowledge, with fast escalation.
Inspect role boundaryMarketing Assistant
Calendars, drafts, approvals and performance summaries coordinated in one rhythm.
Inspect role boundaryLegal Operations Assistant
Contract intake, dates and obligations organised — without ever implying legal advice.
Inspect role boundaryEvery role works inside a reporting line, not instead of one.
An AI employee is never deployed on its own. Each role is configured with a reporting line to a named human owner, a permission set that defines what it can see and prepare, and an escalation path for anything outside its boundary.
Every action the role takes is written to an activity record: what triggered it, what it prepared, what a person decided, and what evidence was retained. Supervision is not a promise — it is a property of the system.
- A reporting line to a named human owner, not a dashboard nobody reads
- Permissions scoped to the role — access is granted, not assumed
- An escalation path for sensitive, ambiguous or out-of-scope cases
- An activity record behind every prepared item and every decision
Bring us one operational workflow.
We will look at how the work moves today, where authority should sit, and what a sensible first phase could look like.