Ask a general AI model what permits a site needs and it will answer confidently, whether or not it actually knows. That's the specific failure mode Civil Proposal is built to avoid. The AI in Civil Proposal doesn't guess from training data about what a site in Kelowna or Thunder Bay probably requires. It reads real municipal, provincial and federal datasets for that specific address first, and only then uses an LLM to organize what it found into a scoping report a civil engineer can actually use.
That ordering, data first, LLM second, is the whole engineering decision behind the product.
Why the order matters
A large language model is good at reading and structuring information. It is not a source of truth about a specific municipality's stormwater bylaw or a specific province's permitting timeline. Civil Proposal's pipeline queries live datasets across Canadian municipalities, provinces and territories for the site in question, then hands that retrieved data, not a general assumption, to the LLM to turn into a structured report. If the pipeline can't retrieve a verified figure for something, the report says so explicitly rather than letting the model fill the gap with a plausible-sounding guess.
That's the difference between an AI tool that sounds confident and one that's actually accurate for a specific site.
What comes back: a six-part scoping report, sourced line by line
Enter a site address, and Civil Proposal returns:
- Risk flags specific to that site
- A planning-level budget range
- Permit context and timelines
- Scope assumptions and exclusions
- Site and servicing context
- Reference standards that apply
Every figure carries one of four labels: Sourced (pulled from a real record for that site), Published standard (a known baseline, not site-verified), Confirm with the authority (no reliable data found, flagged rather than guessed), or Source not available. Each labelled figure also carries the date it was retrieved, so a firm knows exactly how current the data behind their bid actually is.
What this actually saves an engineer
Civil firms typically lose four to six hours of unbilled time per pursuit just researching what applies to a site, often with a day or two before the bid is due. Civil Proposal turns that research phase into a report available in a fraction of the time, built from the same public data an engineer would otherwise track down municipality by municipality, department by department. From there, a firm develops scope and fee inside the tool and exports a structured proposal draft, with a verification checklist and an audit-trail appendix carrying every figure's source and retrieval date.
The job doesn't change. The unpaid research grind that used to sit in front of it does.
Where the data actually comes from
Civil Proposal's coverage runs on two layers: 13 provincial and territorial baselines — all 10 provinces and all 3 territories in Canada — giving every report a jurisdiction-appropriate starting point regardless of where the site is, and 240 municipalities with configured GIS data, where the report goes deeper into site-specific zoning, servicing and permitting detail. A site inside one of those 240 municipalities gets more Sourced, site-verified figures; a site outside that set falls back to the provincial or territorial baseline and shows more Confirm-with-the-authority flags. That's the pipeline being honest about the limits of its own data, not a weakness it hides.
What it's built for, and what it isn't
Civil Proposal is purpose-built for civil engineering site scoping specifically — permitting, servicing, stormwater, zoning, traffic. It's not a tool for electrical, structural, or mechanical engineering work, and we're not claiming it is.
FAQ
Does Civil Proposal's AI use real data, or does it generate answers from a general model?
It queries real municipal, provincial and federal datasets for the specific site address first. The LLM's job is to organize that retrieved data into a report, not to generate answers from general training knowledge.
How much research time does Civil Proposal save on a typical bid?
Civil firms commonly spend four to six hours of unbilled research per pursuit gathering the same information Civil Proposal returns from a site address in a fraction of that time.
What happens when Civil Proposal can't find verified data for something?
It labels that figure "Confirm with the authority" or "Source not available" rather than presenting a guess as a verified fact.
What data coverage does Civil Proposal have across Canada?
Provincial and territorial baselines for all 13 jurisdictions in Canada, plus deeper, site-specific GIS data for 240 configured municipalities.
Try it
If you want to see what a scoping report looks like for a real site, Civil Proposal has a free plan. We're based in Kelowna and Civil Proposal works with civil engineering firms across Canada.