I’m a software engineer by day, but civic tech is something I care deeply about. So I spent some time this week, half-watching The Pitt, building an interactive tool to trace Ontario’s $244 billion budget as deep as the public data allows.

You can drill from the top level down into the things a budget is actually made of: individual school boards (all 72, with verified allocations), named hospital capital projects, transit megaprojects with independent cost tracking, and social programs. Revenue comes in at $231.9B against $244.2B in spending, and from there you can keep clicking down until you hit the limit of what’s public.

Every number is tagged for honesty

The thing I care most about isn’t the visualization. It’s that every single number is honestly labeled with where it came from and how much you should trust it:

• 🟢 Verified: a published government figure you can trace to a source.
• 🟠 Black box: the government tracks this internally but doesn’t share it publicly.
• 🟡 Doesn’t exist in this form: the public data simply isn’t broken out this way; where I’ve approximated, it’s flagged as an estimate.
• 🔴 Classified: genuinely restricted.

Most budget dashboards quietly paper over the gaps. This one refuses to. If a number is a guess, it says so. If the province has the data and won’t release it, the tool calls that out by name.

The findings

When you tag everything honestly, the shape of the problem becomes obvious:

• 🟢 29% verified from published sources
• 🟠 34% black box, tracked internally, not shared publicly
• 🟡 35% the data doesn’t exist in this form
• 🔴 2% classified

The black-box band is the headline: about $162 billion the province tracks in its own systems but doesn’t open to the public. Add the third that doesn’t exist in any public form, and most of the budget can’t actually be followed to ground. That’s a transparency problem. It’s also an opportunity.

Why this is suddenly possible

AI makes a new kind of civic analysis possible. Pulling this together meant synthesizing 72 school-board PDFs, 388 municipal grant tables, auditor general reports, individual agency financial statements, and ministry expenditure estimates into one coherent, interrogable picture, and flagging every gap along the way. That’s a volume and heterogeneity of source material that would have made this a multi-person, multi-month project a couple of years ago. That one person can now assemble it, with every gap flagged, is the actual story.

That shift matters more than the tool itself. Government data isn’t hidden because it’s technically hard to publish. It’s hidden because nobody has built the infrastructure to aggregate, question, and present it at scale. That’s changing. I think the next wave of civic-engagement tools will be built on exactly this capability: taking the black boxes and making them visible.

Have a dig through it yourself, trace the $244B here, and tell me what you find.