Introducing Pulse Drafting - citation-verified AI responses for parliamentary offices

Today we are introducing Pulse Drafting, the AI-powered response generation engine at the heart of Public Pulse. It represents a fundamentally different approach to AI in government: every draft is grounded in your documents, every claim is cited, and every response goes through human approval before it is sent.

Grounded in your knowledge, not general data

Unlike general-purpose AI tools that generate responses from their training data, Pulse Drafting works exclusively from your office's knowledge base. This includes your approved talking points, past correspondence, Hansard contributions, policy documents, and ministerial briefs. The AI cannot invent facts or policy positions because it has no access to anything beyond your verified sources.

Inline citations on every claim

Every factual statement in a Pulse draft is linked to its source document with an inline citation. Staff can click any citation to see the original passage, verify accuracy, and understand the context. This is not a summary or a paraphrase - it is a direct reference to the document that supports the claim.

Your office voice

During onboarding, Pulse analyses your past correspondence to learn your office's writing style: tone, structure, level of formality, and common phrases. Every draft maintains this voice consistently, whether it is a response to a constituent enquiry, a ministerial brief, or a social media reply.

How it works

When a new enquiry arrives, the Pulse Engine classifies it, retrieves relevant context from your knowledge base using Graph RAG, and generates a draft response. The draft appears in your inbox with citations highlighted, ready for review. Your team can edit, approve, or request changes through the built-in approval workflow.

The entire process takes seconds, not hours. But the human approval step ensures that no response is ever sent without a person reviewing it first.

Security by design

Pulse Drafting runs entirely within our Australian-hosted infrastructure. No constituent data is sent to third-party AI providers. No data is used to train models. Every draft, edit, and approval is logged in an immutable audit trail. The system is ISO 27001 certified and CASA Tier 2 assessed.