How to Turn a Redlined Proposal Into a Change Order With AI

The problem and who this is for

This workflow is for freelancers and consultants who already sent a proposal, got it back with edits, and now need to stop silent scope growth before work starts or before the project keeps expanding. The danger is not the markup itself. The danger is accepting new work through comments, tracked changes, or casual email phrasing without pricing and timing it separately.

A change order is not overkill. It is how you keep the client relationship calm while keeping the business side clean.

Prerequisites

You need the original proposal, the marked-up or commented version, and any email or chat message that explains what the client wants changed. You also need your current price and timing assumptions so the tool can describe impact instead of pretending there is none.

ChatGPT is the simplest primary tool here because it supports file uploads and document analysis in a standard chat workflow. Claude and Gemini are both strong alternatives if that better matches your stack.

How to capture and gather the source material

1. If the proposal is already digital, export the clean original as PDF.
2. Export the marked-up version as PDF if comments or tracked changes are inside Google Docs or Word.
3. If the client sent handwritten notes on a printed copy, scan the pages with your phone's document scanner, Notes app, Adobe Scan, Microsoft Lens, or any scanner you already use.
4. Save the client email that explains the edits. If it is short, paste it into a text file so you can upload it with the proposal.
5. Before upload, rename files clearly. Good examples are proposal-original.pdf, proposal-marked-up.pdf, and client-change-email.txt.

The workflow

1. Upload the original proposal and the changed version to ChatGPT.
2. Ask it to compare the two and list every change in plain language.
3. Add the client's explanation email if the reason for the change matters.
4. Have the tool classify each change as clarification, added scope, removed scope, or timing change.
5. Ask for a short change-order draft that does three things: names the change, explains impact, and gives the client a clean approval path.
6. Copy the result into your own template and replace any placeholder price or timeline language with your real numbers.
7. Send the change order before doing the extra work.

Primary tool instructions: ChatGPT

1. Start a new chat for the specific proposal.
2. Upload the original proposal and the marked-up version.
3. Ask for a side-by-side change summary first. This catches bad OCR, missing pages, and misunderstood comments before you draft anything client-facing.
4. Once the change summary looks right, ask for the change order.
5. Keep the output short. A good change order is easier to approve when it is precise and not padded with sales language.

Alternative tool instructions

Claude

Claude is a strong fallback when the proposal is long and the requested edits are buried in a large document. Upload the files, ask for a change table first, then ask for a clean client-ready change order.

Gemini

Gemini works well if the proposal already lives in your Google workflow. Upload the PDF or the scanned pages and use the fallback prompt below. This is especially convenient when the marked-up file came from Google Docs or Drive.

Copy and paste prompt blocks

Primary prompt for ChatGPT

{
  "role": "change-order-drafter",
  "goal": "Turn the uploaded redlined proposal and client comments into a concise change order summary.",
  "inputs": {
    "original_scope": "Use the uploaded proposal or statement of work.",
    "client_changes": "Use the markup, comments, or added-request email.",
    "current_price_and_timeline": "Optional but helpful."
  },
  "instructions": [
    "List each requested change separately.",
    "Classify each item as scope increase, clarification, deletion, or timeline change.",
    "Draft a change order with these sections: Change Summary, Impact on Deliverables, Impact on Timeline, Price Impact, Assumptions, Approval Line.",
    "Use neutral language. Do not sound defensive.",
    "Do not bury new work inside the old scope."
  ],
  "output_format": {
    "style": "short, professional, client-facing",
    "sections": [
      "Change Summary",
      "Impact on Deliverables",
      "Impact on Timeline",
      "Price Impact",
      "Assumptions",
      "Approval Line"
    ]
  }
}

Fallback comparison prompt for Claude or Gemini

{
  "role": "scope-difference-analyzer",
  "goal": "Compare the original proposal to the client's requested edits and show what changed.",
  "inputs": {
    "original_text": "Paste the original scope.",
    "changed_text_or_notes": "Paste the marked-up text or comment summary."
  },
  "instructions": [
    "Create a table with Original Item, Requested Change, Type of Change, and Suggested Treatment.",
    "Flag anything that should move to a separate phase or optional add-on.",
    "Write a short client-ready summary after the table."
  ],
  "output_format": {
    "sections": [
      "Change Table",
      "Recommended Change Order Summary",
      "Open Questions"
    ]
  }
}

Quality checks

• Every new deliverable should appear in the change order as new work, not as a silent rewrite of the old scope.
• Timing impact should be stated even if the change is small.
• Pricing impact should be explicit, even if the amount is zero.
• The approval line should make it obvious what the client is agreeing to.

Common failure modes and fixes

The tool collapses several changes into one paragraph

Tell it to list each requested change as a separate row before drafting the summary.

The scanned markup is messy

Use a document scan rather than a normal photo. Crop edges and retake blurry pages.

The client phrased the request as a question

Treat it as an unapproved possible change until there is written approval. Ask the tool to keep it in Open Questions if needed.

The change feels too small for a formal change order

Use a short version. One page is enough. The goal is clarity, not bureaucracy.

Sources Checked

• https://help.openai.com/en/articles/8555545-uploading-files-with-advanced-data-analysis-in-chatgpt (accessed 2026-03-24)
• https://support.anthropic.com/en/articles/8241126-what-kinds-of-documents-can-i-upload-to-claude-ai (accessed 2026-03-24)
• https://support.google.com/gemini/answer/14903178 (accessed 2026-03-24)

Quarterly Refresh Flag

Review by 2026-06-22. Recheck document upload support and any plan-based limits in ChatGPT, Claude, and Gemini before updating this workflow.