Federal budget analysis, published on budget day.

Stephen Taylor’s AI Public Affairs Toolset reads what governments publish and what is said about it — budgets, legislation, Hansard and committee transcripts, regulatory filings, news coverage, press releases and broadcast interviews — and turns it into structured analysis, audience-specific briefings and publication-ready pages faster than a traditional research team can.

It is not a chatbot or a generic AI wrapper. It is a purpose-built toolset for organisations that need to understand and respond to government action fast. This website is a live demonstration.

Every page on this site is generated by AI agents working from the official budget documents published by the Government of Canada — the budget plan, its annexes and the accompanying tables. The agents extract the text and tables, normalise the numbers, then write sector, audience and stakeholder analysis under one strict rule set: no invented figures, units preserved, fact kept separate from interpretation, and every claim traced to its source.

Sector, audience and stakeholder pages also score what groups asked for before the budget against what it delivered. Those asks are compiled from House of Commons committee hearings, news coverage, press releases and television interviews in the weeks before tabling.

Before anything is published, the figures on each page are checked against the source documents in an automated numeric QA pass, and the copy is reviewed for editorial consistency. Where a number cannot be found in the documents, the analysis says so rather than guessing. Always verify against the official documents before relying on a figure.

federalbudget.ca is made by Stephen Taylor. The site exists to show what his AI Public Affairs Toolset can do with the most consequential document the federal government publishes each year — and to put that analysis in front of the people it affects while the budget is still news.

Here is what you will find on the site.