How to Compare Contractor Repair Estimates Side by Side With AI

Contractor estimates are hard to compare when the vendors use different language, different line-item depth, and different assumptions. One quote might look cheaper only because it excludes hauling, prep, patching, or finish work. This workflow is for property managers, owner-operators, and renovation coordinators who need a faster apples-to-apples review before they call vendors back.

Warning: This workflow creates drafts, summaries, checklists, and briefing documents for review. Double-check unit numbers, dates, conditions, vendor scope, resident details, owner details, and any promised next steps before you use any output. Do not use AI output as the final signed inspection record, an official notice, a legal document, or the only record of property condition.

Problem this solves and who it is for

Use this when you have two or three real estimates for the same repair or turnover project and need a plain-English memo that shows what is actually different. The goal is not to let AI pick the winner. The goal is to surface scope gaps, missing assumptions, and smart clarification questions before you decide.

Prerequisites

Two or three estimates in PDF or document format
A desktop web app workflow for the cleanest comparison
A Claude.ai account or another file-friendly long-document chat tool
Enough project context to know whether all vendors were pricing the same request

How to capture or gather the source material

Before you upload anything, normalize the estimates as much as possible:

save each estimate as a separate PDF
rename them clearly, such as vendor-a-paint-estimate.pdf
remove unrelated pages if they do not affect scope
add a short one-page note that explains the project and any must-have requirements all vendors were supposed to price

If one vendor sent a photo in the body of an email and another sent a polished PDF, move both into a consistent document set first. The cleaner the input set, the cleaner the comparison memo.

Step-by-step workflow

1. Define the comparison question before you upload the files

Write down what the estimates are supposed to cover. For example:

full repaint of a two-bedroom unit
bathroom subfloor repair and finish replacement
make-ready cleaning plus trash-out
flooring replacement in living room and hall

If you do not define the comparison question first, the model will still compare the files, but the memo may not tell you whether the vendors were even bidding the same job.

2. Upload the estimates to your comparison tool

Claude.ai is a strong fit for this because it supports PDF and document uploads and is good at side-by-side document comparison. It supports PDF, DOCX, CSV, TXT, and other document types, with chat uploads up to 20 files per chat and 30 MB per file. For PDFs under 100 pages on supported Claude models, visual elements can also be analyzed.

Desktop primary path

Open Claude.ai in your browser.
Upload the estimates and your one-page project note.
Paste the JSON prompt below.
Ask for a comparison memo, not a recommendation.

Mobile support path

If you already have the PDFs saved on your phone, you can upload them from the mobile app or mobile web flow you normally use for Claude. Desktop is still the better path because estimate comparison is easier when you can inspect the PDFs and the memo side by side.

{
  "task": "Compare these repair estimates side by side for a property operations decision memo.",
  "project_context": {
    "project_name": "[short name of repair or turnover job]",
    "location": "[property and unit if relevant]",
    "decision_goal": "understand scope differences before vendor follow-up"
  },
  "comparison_rules": [
    "Use only the uploaded estimates and project note.",
    "Identify what appears in all estimates.",
    "Identify what appears in only one estimate.",
    "Call out likely omissions, assumptions, allowances, and unclear scope language.",
    "Do not recommend a vendor based on price alone.",
    "Do not invent scope that is not stated."
  ],
  "output_request": "Return a plain-English comparison memo with sections for Common Scope, Key Differences, Possible Omissions, Questions to Clarify, and a short Summary for decision-makers."
}

3. Turn the memo into an apples-to-apples call list

Once the comparison memo is grounded, ask for a shorter follow-up list you can actually use when calling vendors back.

{
  "task": "Turn the comparison memo into a vendor clarification list.",
  "requirements": [
    "Group questions by vendor.",
    "Keep each question short and neutral.",
    "Focus on scope clarity, exclusions, schedule assumptions, prep work, finish work, hauling, and cleanup when relevant.",
    "Do not ask questions that are already clearly answered in the estimate."
  ],
  "output_request": "Return a short follow-up question list for each vendor."
}

4. Add the non-document context yourself

AI can compare the paperwork. It cannot know your whole decision context. Before you decide, add:

who has actually done work for you before
who can start soon enough
who is licensed or insured for the job if required
whether one estimate reflects a different quality level
whether access conditions were explained consistently

5. Save the memo with the estimate set

The final memo is most useful when stored with the original estimates. That way anyone reviewing the decision later can see the source files, not just the summary.

Tool-specific instructions

Use Claude or another file-friendly comparison tool on desktop for the cleanest review. Claude's current help documentation confirms support for PDFs and other common document types, and notes that supported Claude models can analyze both text and visual elements in PDFs under 100 pages. Even so, do not assume the model caught every exclusion or allowance. This workflow still needs a human scope review.

Quality checks

Before you rely on the memo, verify:

the estimates actually cover the same job
the memo distinguishes common scope from unique scope
obvious exclusions are called out
the follow-up list is neutral and concrete
no vendor is treated as more complete simply because the PDF is prettier

Common failure modes and fixes

The memo compares formatting more than scope
Add a one-page project note and rerun the comparison with a clearer comparison question.

One estimate is much less detailed than the others
Ask for a section specifically called Missing or Unclear Scope.

The model starts leaning toward the cheapest estimate
Tell it to compare completeness and assumptions, not price ranking.

An estimate includes photos or diagrams that matter
Use PDF versions and keep them under the visual-analysis threshold when possible.

The follow-up list is too long
Ask for the top five clarification questions per vendor only.

Sources Checked

Anthropic Help Center, "What kinds of documents can I upload to Claude.ai?" Accessed 2026-03-21. https://support.anthropic.com/en/articles/8241126-what-kinds-of-documents-can-i-upload-to-claude-ai
Anthropic Help Center, "What kinds of images can I upload to Claude.ai?" Accessed 2026-03-21. https://support.anthropic.com/en/articles/9002500-what-kinds-of-images-can-i-upload-to-claude-ai
Unsplash License. Accessed 2026-03-21. https://unsplash.com/license
Unsplash Help Center, "Can I use Unsplash images for personal or commercial projects?" Accessed 2026-03-21. https://help.unsplash.com/en/articles/2612315-can-i-use-unsplash-images-for-personal-or-commercial-projects

Quarterly Refresh Flag

Review by 2026-06-19 to confirm Claude.ai supported document types, PDF visual-analysis behavior, and current file-size and file-count guidance.