Niyatna Docs

Niyatna Agents & Identities

Scoped workforce units with roles, permissions, tools, memory, and proof duties.


Niyatna Agents are accountable workforce units inside the company system.

They are not generic chat assistants and they are not replacements for company structure. Each agent receives a defined role, tool scope, memory scope, proof duty, and report-back path.


Agent identity model

A Niyatna Agent should be described by:

  • Mission: What company outcome it supports.
  • Role boundary: Which department, workflow, or operating surface it belongs to.
  • Permission scope: What it can read, write, schedule, or escalate.
  • Tools: Which company systems it can use.
  • Memory access: Which SOPs, decisions, and proof history it can retrieve.
  • Proof duty: What evidence it must return before work is accepted.
  • Report-back path: Who receives the final result.

Example agent roles

AgentPurposeBoundary
Operations AgentTurns recurring operations requests into assigned work and follow-up.Tasks, schedules, SOPs, and approved operations tools.
Finance AgentReviews finance submissions against company standards and attached rules.Finance workspace, read-only records, and approval-gated report-back.
Proof AgentChecks whether completed work matches the original intent and evidence standard.Proof artifacts, evaluation notes, and memory write-back.
Research AgentCollects source-backed context for a scoped decision or workflow.Approved sources and clearly cited outputs.

Agent rule

An agent is useful only when its authority is legible. If a company cannot see what the agent was asked to do, what it touched, and what proof it returned, the work is not complete.