Can ChatGPT check my contractor quote?
What a general chatbot does well with a renovation quote, where it goes wrong on Canadian prices, and what an independent reference range adds. A fair comparison.
Updated 2026-09-02
The short answer
Yes, for the parts that are language: explaining terms, spotting a missing item, drafting questions for the contractor. No, for the part that matters most: telling you whether $24,000 is a fair price for your roof in your city this year. A chatbot answers that from memory, with no source, no local data, and a different number each time you ask.
What a chatbot does well
Paste a quote into any capable chatbot and it will translate "ice and water shield at eaves and valleys" into plain English, notice the quote never mentions a permit, and write five sharp questions to send back. It can also explain why quotes vary and what a warranty clause means. Use it for that.
- Explaining materials, brands and trade terms
- Listing what a quote should include for a given trade
- Drafting questions and negotiation messages
- Summarizing a long, messy PDF
Where it goes wrong
Ask the same chatbot what the roof should cost and the problems start. The range comes from training data with no date and no city, is often US pricing quietly converted, and shifts between conversations. It cannot cite the Job Bank wage for roofers in your region, the permit fee schedule for your city, or what shingles cost at a Canadian retailer this month. It will present a confident number anyway.
- No local data: Vancouver labour is 20 to 30 percent above the national average, and a national figure hides that.
- No date: material prices moved sharply between 2021 and 2026; a memorized range does not say which year it is from.
- No consistency: two runs give two ranges, so you cannot rely on it to compare three quotes on equal terms.
- No audit trail: you cannot see which sources produced the number, so you cannot judge it.
What a reference range adds
QuoteSaver reads the quote the same way, then prices the same scope from a published reference model: 48 cited Canadian sources, a regional index built from Job Bank median trade wages and local price premiums, and municipal permit schedules. The report shows the range, the assumptions, the sources, and the gap in dollars. Run it again and the number is the same.
Comparing quotes is where the difference is largest
Three quotes rarely describe the same job. One removes two layers, one includes premium shingles, one carries a project-management fee. A chatbot will tell you which total is lowest. A reference model values each extra, removes it, and shows what each contractor charges for the same work. That normalized price, plus warranty and references, is how you choose.
A fair way to use both
Use a chatbot to understand the quote and prepare your questions. Use QuoteSaver to find out what it should cost in your city and to compare quotes like for like. Then ask the contractor about the specific gaps the report found.
Frequently asked questions
- Is QuoteSaver just ChatGPT with a wrapper?
- No. A language model reads the quote and extracts the line items, but the price range comes from a separate reference model with cited Canadian sources and a regional index. The model never invents the number.
- Will my quote be used to train an AI?
- No. The model reads it once to extract the numbers and does not train on it. Your quote is stored in Canada, and it joins the anonymized price index only if you opt in.
- How accurate is the range?
- It is a source-backed estimate with a stated band, from ±14 percent for roofing to ±25 percent for kitchens, and it has not yet been calibrated against observed quotes. Every report says so, and the methodology page lists every source.
- Can I paste a chatbot's summary into QuoteSaver?
- You can, but paste the original quote instead. The report depends on the exact line items, amounts and scope facts as the contractor wrote them.