Skip to content
RESPONSIBLE AI

AI that people can trust is built, not declared.

Responsibility at AIGrama is not a compliance page — it is how engagements are scoped, staffed, and measured. These are the commitments every piece of work carries.

01
Human oversight, always

AI systems we design keep people in the loop for consequential decisions. In citizen-facing work, no automation ships without human review and a clear escalation path.

02
Data minimisation and purpose limitation

We collect and process only the data a workflow needs, for the purpose it was gathered — designed to align with the DPDP Act, 2023 and the client's own governance.

03
Evaluation before scale

Success criteria are agreed before a system is built, measured in a bounded pilot, and reviewed before anything scales. A demo is not adoption.

04
Capability boundaries, stated plainly

We describe what a system can and cannot do — to buyers, users, and governors. Our own scope follows the same rule: AI is current practice, Robotics is training, Quantum is a horizon.

05
People are trained as part of the work

Every system ships with its users and governors enabled — training, operating notes, and an improvement backlog are deliverables, not extras.

06
Documented, auditable handover

Decisions, data flows, and ownership are documented so the work stands up to audit — and so the capability belongs to you, not to us.