---
title: How do I build a connection reactions register from the design drawings and flag reactions to RFI with AI? | AEC Agents
url: https://aecagents.si/use-cases/connection-reactions-register-rfi-flags
description: When connection design is delegated, the reactions come off the engineer of record's plans, schedules and general notes in a dozen formats: a beam end shear table on one sheet, a typical-reaction note on another, axial forces tagged on a framing plan and sometimes out-of-plane shears nobody expected. Detailers and the connection engineer lose days just collecting them, and the unusual ones surface late, after connections are already sized. Have Claude or ChatGPT read the design drawings into one register with a row per member end, and have Copilot in Excel or Gemini in Google Sheets flag the rows that match rules you set for an RFI: missing reactions, values that disagree between sheets, and force components you decided to question. Whether a reaction is reasonable, and how the connection is designed, stays with the connection engineer and the engineer of record.
last_modified: 2026-11-11
source: AEC Agents (free, open workflows; not engineering advice)
---

[AI assists](https://aecagents.si/use-cases/connection-reactions-register-rfi-flags#honesty-labels)Workflow guide · not independently tested

# How do I build a connection reactions register from the design drawings and flag reactions to RFI with AI?

When connection design is delegated, the reactions come off the engineer of record's plans, schedules and general notes in a dozen formats: a beam end shear table on one sheet, a typical-reaction note on another, axial forces tagged on a framing plan and sometimes out-of-plane shears nobody expected. Detailers and the connection engineer lose days just collecting them, and the unusual ones surface late, after connections are already sized. Have Claude or ChatGPT read the design drawings into one register with a row per member end, and have Copilot in Excel or Gemini in Google Sheets flag the rows that match rules you set for an RFI: missing reactions, values that disagree between sheets, and force components you decided to question. Whether a reaction is reasonable, and how the connection is designed, stays with the connection engineer and the engineer of record.

[Fabricator](https://aecagents.si/roles/fabricator) · Fabrication · Claude · ChatGPT · Microsoft Excel · Microsoft Copilot · Google Sheets

Content updated 11 November 2026 · Tool links listed below; vendor features and pricing change, so confirm with the vendor

## Short answer

Use Claude or ChatGPT, then Copilot in Excel or Gemini in Google Sheets: Claude or ChatGPT reads the design drawings into a register with one row per member end and lists conflicts between sheets; Copilot or Gemini applies your own RFI rules and groups flagged reactions by sheet and member type. When connection design is delegated, the reactions come off the engineer of record's plans, schedules and general notes in a dozen formats: a beam end shear table on one sheet, a typical-reaction note on another, axial forces tagged on a framing plan and sometimes out-of-plane shears nobody expected. Detailers and the connection engineer lose days just collecting them, and the unusual ones surface late, after connections are already sized. Have Claude or ChatGPT read the design drawings into one register with a row per member end, and have Copilot in Excel or Gemini in Google Sheets flag the rows that match rules you set for an RFI: missing reactions, values that disagree between sheets, and force components you decided to question. Whether a reaction is reasonable, and how the connection is designed, stays with the connection engineer and the engineer of record. A person still checks: every flagged row against the drawings, and every RFI, with the connection engineer before it goes to the engineer of record.

## AI tie-in

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

Claude or ChatGPT reads the design drawings into a register with one row per member end and lists conflicts between sheets; Copilot or Gemini applies your own RFI rules and groups flagged reactions by sheet and member type.

### Prompt 1

```
Read these structural design drawings [upload]. Build a table with one row per member end: sheet, revision, location, member as written, end, each force component as shown with units, and whether it is tabulated or from a typical note. Copy values exactly and write NOT SHOWN for gaps. Then list members whose values differ between sheets. Don't judge whether any reaction is correct.
```

### Prompt 2

```
Here are the headers and a few sample rows of my Reactions and Rules tabs [paste]. Write Excel formulas (or a step-by-step Copilot plan) that flag NOT SHOWN components, cross-sheet conflicts and rows meeting my rules, and group flagged rows by sheet and member type for one RFI per group. Use only my data and don't design connections.
```

**Human check:** Every flagged row against the drawings, and every RFI, with the connection engineer before it goes to the engineer of record.

## The problem

Build a connection reactions register from the design drawings and flag reactions to RFI with AI. This is a common, time-consuming task for a fabricator.

## What this workflow does

When connection design is delegated, the reactions come off the engineer of record's plans, schedules and general notes in a dozen formats: a beam end shear table on one sheet, a typical-reaction note on another, axial forces tagged on a framing plan and sometimes out-of-plane shears nobody expected. Detailers and the connection engineer lose days just collecting them, and the unusual ones surface late, after connections are already sized. Have Claude or ChatGPT read the design drawings into one register with a row per member end, and have Copilot in Excel or Gemini in Google Sheets flag the rows that match rules you set for an RFI: missing reactions, values that disagree between sheets, and force components you decided to question. Whether a reaction is reasonable, and how the connection is designed, stays with the connection engineer and 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 register only reflects the drawings you gave it and the rules you typed. A flag means a question to ask, not that the engineer of record is wrong, and a clean row doesn't mean the reaction is complete. Connection design, load interpretation and every RFI that goes out stay with the connection engineer, who checks each row against the drawings.

## 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

- [Claude (Anthropic)](https://www.anthropic.com/claude)
- [ChatGPT](https://chatgpt.com/)
- [Microsoft Excel](https://www.microsoft.com/microsoft-365/excel)
- [Microsoft Copilot](https://www.microsoft.com/microsoft-365/copilot)
- [Google Sheets](https://workspace.google.com/products/sheets/)

## Follow the steps

1. Collect the current structural design drawings, including framing plans, any reaction or connection force schedules, and the general notes that state default reactions or design basis, plus the latest revision list. Use an AI account your company approves for contract documents.
2. Have Claude or ChatGPT read them into a Reactions tab with one row per member end: sheet, revision, grid or location, member mark or size as written, end, and each force component as shown (shear, axial, moment, out-of-plane shear), with the units and whether the value is stated for that member or comes from a typical note. It copies values exactly and writes NOT SHOWN instead of filling a gap.
3. Have it list, on a Conflicts tab, any member that appears on more than one sheet with different values, and any typical note that contradicts a tabulated value. It reports the difference and doesn't pick a winner.
4. Type your RFI rules into a Rules tab yourself: which force components you want questioned, any threshold that triggers a question, and which members must have explicit reactions rather than a typical note.
5. Have Copilot in Excel or Gemini in Google Sheets apply the rules row by row and flag missing reactions, conflicts, values over your thresholds and force components you chose to question, then group the flags by sheet and member type so one RFI can cover many similar ends.
6. Draft the RFI from the grouped flags with the sheet references, have the connection engineer review every question before it goes out, and log the answer back against each row when it returns.

## 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 Reactions tab [paste: sheet, revision, location, member, end, shear, axial, moment, out-of-plane shear, units, source (tabulated or typical note)] and my Rules tab I typed myself [paste: component, threshold, applies to]. Write Excel formulas (or a step-by-step Copilot plan) that flag rows with any NOT SHOWN component, rows that conflict with another sheet for the same member, and rows that meet any of my rules, then build a summary grouping flagged rows by sheet and member type for one RFI per group. Use only my data; do not judge whether a reaction is correct or design any connection.
```

### 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 register only reflects the drawings you gave it and the rules you typed. A flag means a question to ask, not that the engineer of record is wrong, and a clean row doesn't mean the reaction is complete. Connection design, load interpretation and every RFI that goes out stay with the connection engineer, who checks each row against the drawings.
- 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 the design drawings into a register with one row per member end and lists conflicts between sheets; Copilot or Gemini applies your own RFI rules and groups flagged reactions by sheet and member type. A person still checks: Every flagged row against the drawings, and every RFI, with the connection engineer before it goes to the engineer of record.

How do I how do I build a connection reactions register from the design drawings and flag reactions to RFI with AI??

Collect the current structural design drawings, including framing plans, any reaction or connection force schedules, and the general notes that state default reactions or design basis, plus the latest revision list. Use an AI account your company approves for contract documents. Have Claude or ChatGPT read them into a Reactions tab with one row per member end: sheet, revision, grid or location, member mark or size as written, end, and each force component as shown (shear, axial, moment, out-of-plane shear), with the units and whether the value is stated for that member or comes from a typical note. It copies values exactly and writes NOT SHOWN instead of filling a gap. Have it list, on a Conflicts tab, any member that appears on more than one sheet with different values, and any typical note that contradicts a tabulated value. It reports the difference and doesn't pick a winner. 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, Google Sheets. 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.

## Where to go next

- [How do I answer shop questions from drawings and specs with AI document search?](https://aecagents.si/use-cases/fabrication-shop-docs)
- [How do I go from design model to shop drawings and CNC data in detailing software?](https://aecagents.si/use-cases/steel-detailing-model)
- [How do I turn mill test reports into a material heat number traceability log with AI?](https://aecagents.si/use-cases/mtr-heat-number-traceability-log)
- [All Fabricator workflows](https://aecagents.si/roles/fabricator)
- [All Design calcs and analysis workflows](https://aecagents.si/tasks/design-calcs)

## 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.

- [Claude (Anthropic) (vendor site)](https://www.anthropic.com/claude)
- [ChatGPT (vendor site)](https://chatgpt.com/)
- [Microsoft Excel (vendor site)](https://www.microsoft.com/microsoft-365/excel)
- [Microsoft Copilot (vendor site)](https://www.microsoft.com/microsoft-365/copilot)
- [Google Sheets (vendor site)](https://workspace.google.com/products/sheets/)

## Put it to work

Start with a small sample and keep the original record open while you verify the result.

## Know what does the work.

Agent executes
:   An AI agent calls connected tools to carry out bounded actions. You set permissions and review the result.

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

Conventional automation
:   Rules or specialist software perform the task. No AI agent is required.
