How do I pull actual start and finish dates from daily reports into the schedule update with AI?

Short answer

Use Claude or ChatGPT, then Copilot in Excel or Gemini in Google Sheets: Claude or ChatGPT reads daily reports into dated work events and proposes which schedule activity each belongs to; Copilot or Gemini finds first and last work dates per activity and flags actual dates in the update that the reports don't support. Monthly updates often get actual dates from memory or a quick call with the superintendent, and the daily reports that actually say when work started and finished sit unread in a folder. Wrong actuals push through the whole update: activities show started when they weren't, or finished weeks after the crew left. Have Claude or ChatGPT read the month's daily reports into a list of work events with dates, and have Copilot in Excel or Gemini in Google Sheets match those events to your activity list and flag differences with the actuals you plan to enter. You still decide every date that goes into the schedule. A person still checks: every proposed activity match and every actual date with the superintendent before it goes into the update.

AI tie-in

Tool: Claude or ChatGPT, then Copilot in Excel or Gemini in Google Sheets

Claude or ChatGPT reads daily reports into dated work events and proposes which schedule activity each belongs to; Copilot or Gemini finds first and last work dates per activity and flags actual dates in the update that the reports don't support.

Prompt 1

Read these daily reports [upload]. Build a table with one row per work item: date, trade, area as written, work as written, crew size and any note such as started, completed, stopped or remobilized. Copy text exactly and write NOT STATED instead of guessing. Then propose the matching activity from this activity list [paste] with a confidence word (clear, likely, unsure).

Prompt 2

Here are the headers and a few sample rows of my Work and Activities tabs [paste]. Write Excel formulas (or a step-by-step Copilot plan) that find the first and last report date per activity and flag actuals that differ by more than [days] days, report work with no actual start, finishes entered while work continues and gaps without work. Use only my data and don't change dates or judge delay responsibility.

Human check: Every proposed activity match and every actual date with the superintendent before it goes into the update.

The problem

Pull actual start and finish dates from daily reports into the schedule update with AI. This is a common, time-consuming task for a scheduler.

What this workflow does

Monthly updates often get actual dates from memory or a quick call with the superintendent, and the daily reports that actually say when work started and finished sit unread in a folder. Wrong actuals push through the whole update: activities show started when they weren't, or finished weeks after the crew left. Have Claude or ChatGPT read the month's daily reports into a list of work events with dates, and have Copilot in Excel or Gemini in Google Sheets match those events to your activity list and flag differences with the actuals you plan to enter. You still decide every date that goes into the schedule.

How it works: AI drafts or analyzes. A person operates the tools, checks the work, and takes action.

Your checkpoint: The comparison only shows what the daily reports say against the dates in your update. Daily reports can be vague about area or which activity the work belongs to, so every match is a proposal. Actual dates, percent complete and anything used in a delay or time-extension claim are decided by the scheduler and superintendent, with the project manager.

Before you start

Access to the named tool; confirm license, version and cost with the vendor.
A small, non-confidential sample to try the workflow on first.
A qualified person to review the output before it is used.

Tools and access

Follow the steps

Collect the period's daily reports as PDFs or an export from your field app, and export the activity list from your scheduling software: activity ID, name, area, planned start and finish, current actual start and finish, percent complete.
Have Claude or ChatGPT read the daily reports into a Work tab: date, trade, area or location as written, work described as written, crew size and any note like started, completed, stopped or remobilized. It copies text exactly and writes NOT STATED instead of guessing an area.
Have it propose a match from each work line to an activity ID using area and description, with a confidence word (clear, likely, unsure). You review the unsure ones and fix the mapping.
Have Copilot in Excel or Gemini in Google Sheets compute, per activity, the first and last report date with matching work, and compare them to the actual dates in your update.
Flag activities marked started with no work in the reports, work in the reports on activities still shown not started, finishes entered while reports show work continuing, and gaps where a crew left and came back.
Walk the flagged list with the superintendent, decide the dates, and enter them yourself. Keep the Work tab with the update as backup for the actuals.

Try a small example

Replace the bracketed fields with your own project details. No tested sample run is published for this workflow yet.

Sample input

Fill in the bracketed [fields] in the text below with your project data.

Starter prompt

Here are the headers and a few sample rows of my Work tab built from daily reports [paste: date, trade, area, work as written, note, matched activity ID] and my Activities tab [paste: activity ID, name, area, planned start, planned finish, actual start, actual finish, percent complete]. Write Excel formulas (or a step-by-step Copilot plan) that find the first and last report date per activity, flag activities with actual dates that don't match the reports by more than [days] days, activities with report work but no actual start, finishes entered while work continues in later reports, and gaps of more than [days] days without work. Use only my data; do not change any dates or judge delay responsibility.

Expected output

Not published yet. Treat any output as a draft and check it against the source documents.

Check the result

Every value, citation or quantity traces back to a source document.
Missing information is marked as missing, not filled in.

Limits and recovery

The comparison only shows what the daily reports say against the dates in your update. Daily reports can be vague about area or which activity the work belongs to, so every match is a proposal. Actual dates, percent complete and anything used in a delay or time-extension claim are decided by the scheduler and superintendent, with the project manager.
Check your organization’s data policy before sharing project files with third-party tools.

Stop or switch when: The source cannot be verified, permissions are missing, or the result contains unsupported facts.

Frequently asked questions

How does AI help with this?

Claude or ChatGPT, then Copilot in Excel or Gemini in Google Sheets: Claude or ChatGPT reads daily reports into dated work events and proposes which schedule activity each belongs to; Copilot or Gemini finds first and last work dates per activity and flags actual dates in the update that the reports don't support. A person still checks: Every proposed activity match and every actual date with the superintendent before it goes into the update.