How do I check field stakeout shots against design coordinates and flag anything outside my tolerance with an AI-written script?

Short answer

Use Claude, ChatGPT or OpenAI Codex: The AI writes a Python script that pairs your design points with the as-staked shots by point ID, computes delta N, delta E, horizontal offset and delta Z, flags pairs outside the tolerances you enter, and lists unmatched and duplicate points in one Excel QA report. Most data collectors can show design and staked values one point at a time, but the job folder often ends up with a raw export and no single record of how every stake compared with design. Export the design points and the stakeout or as-staked shots to CSV, have Claude, ChatGPT or Codex write a short Python script that pairs them by point ID, computes the northing, easting, horizontal and vertical differences, and flags anything outside the tolerances you type in, and save the result as a QA report in the job folder. Pairing rules, tolerances and whether a stake gets reset are yours, not the model's. A person still checks: that both files share the same coordinate system, datum, units and point naming; a sample of pairs against the collector's own stakeout report; and acceptance and any reset by the surveyor in charge.

AI tie-in

Tool: Claude, ChatGPT or OpenAI Codex

The AI writes a Python script that pairs your design points with the as-staked shots by point ID, computes delta N, delta E, horizontal offset and delta Z, flags pairs outside the tolerances you enter, and lists unmatched and duplicate points in one Excel QA report.

Prompt 1

Here are the first 10 lines of my design point file and the first 10 lines of my as-staked export [paste both; generalize descriptions if needed]. Write a commented Python script using pandas that reads files in these formats, pairs each staked shot with its design point by point ID (or by the stake-to field if present), and writes an Excel workbook with: one row per pair showing design N, E, Z, staked N, E, Z, delta N, delta E, horizontal offset, delta Z, and OUT where the horizontal or vertical offset exceeds my tolerances; a tab of design points never staked; a tab of shots with no design match; and a tab of duplicate IDs. Put the tolerances and units in a settings block at the top, leave them blank, and stop with a message if I have not filled them in. Do not pair by nearest point, do not convert units or datums, and do not invent missing values.

Prompt 2

Here is the summary tab from my stakeout QA report [paste]. Add a short section to the script that reports the mean and spread of delta N, delta E and delta Z across all pairs, and lists the points whose offsets differ most from that mean. Describe the pattern in plain words (for example, most points shifted the same direction) and phrase it as something for me to check in the setup, control or design file. Do not say what caused it and do not say whether the stakeout is acceptable.

Human check: That both files share the same coordinate system, datum, units and point naming; a sample of pairs against the collector's own stakeout report; and acceptance and any reset by the surveyor in charge.

The problem

Check field stakeout shots against design coordinates and flag anything outside your tolerance with an AI-written script. This is a common, time-consuming task for a land surveyor.

What this workflow does

Most data collectors can show design and staked values one point at a time, but the job folder often ends up with a raw export and no single record of how every stake compared with design. Export the design points and the stakeout or as-staked shots to CSV, have Claude, ChatGPT or Codex write a short Python script that pairs them by point ID, computes the northing, easting, horizontal and vertical differences, and flags anything outside the tolerances you type in, and save the result as a QA report in the job folder. Pairing rules, tolerances and whether a stake gets reset are yours, not the model's.

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

Your checkpoint: The script compares two files; it does not know whether the control, setup, grid or design file was right. A wrong datum, unit or point naming can make every result wrong. Tolerances come from the job requirements and your own procedures, and the licensed surveyor in charge decides what is acceptable and what gets reset.

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

Export the design points you staked from (point ID, northing, easting, elevation, description) and the as-staked shots from the data collector or office software to CSV. Note the column order (PNEZD or PENZD), units and the coordinate system and vertical datum of each file in a short header note.
Paste the first 10 lines of each file into Claude, ChatGPT or Codex and have it write a commented Python script that reads both, pairs shots to design points by point ID (or by the stake-to point field your collector records), and never pairs by nearest point unless you turn that on.
Type your horizontal and vertical tolerances for this job into the settings block at the top of the script yourself. The script should stop if they are blank, and it must not convert units or datums on its own.
Run it on your own computer. The script writes an Excel report with one row per pair: point ID, design and staked coordinates, delta N, delta E, horizontal offset, delta Z and an OUT flag, plus tabs for unmatched design points, unmatched shots and duplicate IDs.
Look at the summary: if most points are off by about the same amount in the same direction, check the setup, control and grid before blaming the stakes. Check a few rows against the collector's own stakeout report.
Save the report as a PDF or workbook in the job folder with the date, crew and instrument, and rerun the same script on the next stakeout so every job gets the same QA record.

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 first 10 lines of my design point file and the first 10 lines of my as-staked export [paste both; generalize descriptions if needed]. Write a commented Python script using pandas that reads files in these formats, pairs each staked shot with its design point by point ID (or by the stake-to field if present), and writes an Excel workbook with: one row per pair showing design N, E, Z, staked N, E, Z, delta N, delta E, horizontal offset, delta Z, and OUT where the horizontal or vertical offset exceeds my tolerances; a tab of design points never staked; a tab of shots with no design match; and a tab of duplicate IDs. Put the tolerances and units in a settings block at the top, leave them blank, and stop with a message if I have not filled them in. Do not pair by nearest point, do not convert units or datums, and do not invent missing values.

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 script compares two files; it does not know whether the control, setup, grid or design file was right. A wrong datum, unit or point naming can make every result wrong. Tolerances come from the job requirements and your own procedures, and the licensed surveyor in charge decides what is acceptable and what gets reset.
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, ChatGPT or OpenAI Codex: The AI writes a Python script that pairs your design points with the as-staked shots by point ID, computes delta N, delta E, horizontal offset and delta Z, flags pairs outside the tolerances you enter, and lists unmatched and duplicate points in one Excel QA report. A person still checks: That both files share the same coordinate system, datum, units and point naming; a sample of pairs against the collector's own stakeout report; and acceptance and any reset by the surveyor in charge.