Track concrete cylinder break reports in a quality log and flag sets for engineer review with AI

Short answer

Use Claude or ChatGPT, then Copilot in Excel: The AI reads lab strength reports into one row per cylinder (UNREADABLE instead of guesses), then Copilot averages the cylinders in each test, compares tests with the criteria you typed in from the project specification, marks CHECK items, and lists overdue breaks and untested hold cylinders. Strength test reports arrive as PDFs from the lab, one or more per placement, and low results can sit unnoticed until someone rebuilds the log. Have Claude or ChatGPT read each report into a fixed table (sample ID, placement location, mix, cast date, age, individual results), then use Copilot in Excel to average the cylinders that make up each test, line results up against the acceptance criteria you type in from the project specification, and list sets below your entered criteria, overdue breaks and hold cylinders still to be tested. Acceptance and any follow-up are decided by the engineer of record. A person still checks: every transcribed result against the lab report and every criterion against the specification; acceptance and follow-up are decided by the engineer of record.

AI tie-in

Tool: Claude or ChatGPT, then Copilot in Excel

The AI reads lab strength reports into one row per cylinder (UNREADABLE instead of guesses), then Copilot averages the cylinders in each test, compares tests with the criteria you typed in from the project specification, marks CHECK items, and lists overdue breaks and untested hold cylinders.

Prompt 1

Read these concrete compressive strength test reports [upload PDFs]. Build a table with one row per cylinder: file name, page, report number, sample set ID, placement location as printed, mix ID, cast date, test date, age in days, cylinder number, specimen size, measured strength with units, and any notes on fracture type or specimen condition. Copy values exactly as printed. Write UNREADABLE for anything you cannot read and NOT SHOWN for anything missing. Do not judge whether any result is acceptable.

Prompt 2

Here is my cylinder results table [paste], my placement log with planned test ages and cylinder counts [paste], and my criteria tab, typed by me from our project specification and the code edition it references: specified strength per mix, test age, cylinders per test, and the comparison rules to apply [paste]. Do not use any criteria or values from memory. Average the cylinders in each test, apply only my rules, and mark a test CHECK where it falls below one of my criteria, naming the criterion. Never write accepted or rejected. Then list tests past their planned age with no result, hold cylinders not yet tested, and placements with fewer cylinders than planned.

Human check: Every transcribed result against the lab report and every criterion against the specification; acceptance and follow-up are decided by the engineer of record.

The problem

Track concrete cylinder break reports in a quality log and flag sets for engineer review with AI. This is a common, time-consuming task for a inspector / plan reviewer.

What this workflow does

Strength test reports arrive as PDFs from the lab, one or more per placement, and low results can sit unnoticed until someone rebuilds the log. Have Claude or ChatGPT read each report into a fixed table (sample ID, placement location, mix, cast date, age, individual results), then use Copilot in Excel to average the cylinders that make up each test, line results up against the acceptance criteria you type in from the project specification, and list sets below your entered criteria, overdue breaks and hold cylinders still to be tested. Acceptance and any follow-up are decided by the engineer of record.

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

Your checkpoint: The log organizes and compares numbers against criteria you entered; it does not decide acceptance. Acceptance criteria come from the project specification and the adopted code edition, and the engineer of record decides what a low result means and what happens next. Check every transcribed result against the lab report.

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

Keep one log per job with a placement tab (date, location, mix, quantity, ticket numbers, sample set IDs, cylinders made and planned test ages) and a results tab, one row per cylinder.
Upload the lab's strength reports to Claude or ChatGPT and have it fill the results tab exactly as printed: report number, sample set ID, cast date, test date, age, cylinder number, size, measured strength and any fracture or specimen condition notes, with the file name and page. Anything it cannot read is UNREADABLE.
Type the acceptance criteria into a criteria tab yourself, from the project specification and the code edition it references (for example ACI 318 Section 26.12, read in your own licensed copy): specified strength per mix, test age, how many cylinders make up one test, and the rules you are told to check. Tell the model not to supply criteria or values from memory.
Have Copilot in Excel average the cylinders in each test, compute any running averages the criteria tab asks for, and mark each test CHECK where it falls below a criterion you entered, with the criterion it was compared to. It never writes ACCEPTED or REJECTED.
Add a due list: tests whose planned age has passed with no result, hold cylinders not yet broken, and placements with fewer cylinders logged than planned.
Send CHECK items to the engineer of record and the contractor the way the project requires, with the report attached. Have the model draft a factual cover note from the log only; the engineer decides acceptance, cores or other follow-up.

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

Read these concrete compressive strength test reports [upload PDFs]. Build a table with one row per cylinder: file name, page, report number, sample set ID, placement location as printed, mix ID, cast date, test date, age in days, cylinder number, specimen size, measured strength with units, and any notes on fracture type or specimen condition. Copy values exactly as printed. Write UNREADABLE for anything you cannot read and NOT SHOWN for anything missing. Do not judge whether any result is acceptable.

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 log organizes and compares numbers against criteria you entered; it does not decide acceptance. Acceptance criteria come from the project specification and the adopted code edition, and the engineer of record decides what a low result means and what happens next. Check every transcribed result against the lab report.
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: The AI reads lab strength reports into one row per cylinder (UNREADABLE instead of guesses), then Copilot averages the cylinders in each test, compares tests with the criteria you typed in from the project specification, marks CHECK items, and lists overdue breaks and untested hold cylinders. A person still checks: Every transcribed result against the lab report and every criterion against the specification; acceptance and follow-up are decided by the engineer of record.

How do I track concrete cylinder break reports in a quality log and flag sets for engineer review with AI?

Keep one log per job with a placement tab (date, location, mix, quantity, ticket numbers, sample set IDs, cylinders made and planned test ages) and a results tab, one row per cylinder. Upload the lab's strength reports to Claude or ChatGPT and have it fill the results tab exactly as printed: report number, sample set ID, cast date, test date, age, cylinder number, size, measured strength and any fracture or specimen condition notes, with the file name and page. Anything it cannot read is UNREADABLE. Type the acceptance criteria into a criteria tab yourself, from the project specification and the code edition it references (for example ACI 318 Section 26.12, read in your own licensed copy): specified strength per mix, test age, how many cylinders make up one test, and the rules you are told to check. Tell the model not to supply criteria or values from memory. The full step list is above.

Does this use an AI agent or regular software?

It is labeled AI assists: AI drafts or analyzes. A person operates the tools, checks the work, and takes action.

What tools do I need?

Claude, ChatGPT, Microsoft Excel, Microsoft Copilot. Confirm license, version and cost with the vendor, and try it on a small non-confidential sample first.

Has this workflow been tested?

Workflow guide · not independently tested. Written from vendor documentation and practitioner use. Labels apply to the steps described here, not every feature of a product. Links were checked when the page was written.

Evidence and maintenance

Written from vendor documentation and practitioner use. Labels apply to the steps described here, not every feature of a product. Links were checked when the page was written.