Step 2: Connect Google Sheets

Pull interviewer feedback data and later write scores back to Google Sheets.

1. Open Retrieve Feedback Records and select the target spreadsheet and sheet.
2. Credential Required: Connect your googleSheetsOAuth2Api credentials in Retrieve Feedback Records.
3. In Update Score Sheet, set Operation to update and map the fields to {{ $json.Flags }}, {{ $json.Score }}, {{ $json.LLM_JSON }}, and {{ $('Retrieve Feedback Records').item.json.row_number }}.
4. Credential Required: Connect your googleSheetsOAuth2Api credentials in Update Score Sheet.
5. In Append AI Error Log, set Operation to append and choose the error log sheet.
6. Credential Required: Connect your googleSheetsOAuth2Api credentials in Append AI Error Log.

Step 3: Set Up AI Evaluation

Use the AI chain to score feedback quality and validate the model output.

1. Open Evaluate Feedback Quality and keep the text prompt as provided to enforce the JSON-only output.
2. Ensure the input message template includes {{$json["Role"]}}, {{$json["Stage"]}}, and {{$json["Feedback_Text"]}} for context.
3. Open LLM Quality Assessor and set Model to gpt-4o-mini.
4. Credential Required: Connect your azureOpenAiApi credentials in LLM Quality Assessor.
5. Note: LLM Quality Assessor is connected as the language model for Evaluate Feedback Quality—ensure credentials are added to LLM Quality Assessor, not the chain node.
6. In Validate Model Output, keep the condition {{ $json.text }} not equals undefined to route invalid outputs to logging.

Step 4: Parse and Score the AI Output

Convert the AI JSON string into data and compute a weighted score for analysis.

1. In Parse Model JSON, keep the provided jsCode that parses $json["text"] and throws an error if the JSON is invalid.
2. In Compute Weighted Score, keep the weights and scoring logic to generate Score, Flags, LLM_JSON, and VaguePhrasesFormatted.
3. Confirm the node retains row_number, Role, and Stage using references like $item(0).$node["Retrieve Feedback Records"].json.Role.

Step 5: Configure Slack Outputs and Training Routing

Send a summary to Slack and optionally send coaching resources for low scores.

1. Open Post Feedback Summary and keep the text field as provided to format the Slack message.
2. Credential Required: Connect your slackApi credentials in Post Feedback Summary.
3. In Assess Training Need, keep the condition {{$json["Score"]}} less than 50 to trigger coaching.
4. Open Send Coaching Resources and keep the text field as provided for the training recommendation.
5. Credential Required: Connect your slackApi credentials in Send Coaching Resources.
6. Compute Weighted Score outputs to both Post Feedback Summary and Update Score Sheet and Assess Training Need in parallel.

Step 6: Add Error Handling

Log model output issues to a dedicated Google Sheet.

1. From Validate Model Output, ensure the false branch connects to Append AI Error Log.
2. Confirm Append AI Error Log uses Operation append to capture error rows.

Step 7: Test & Activate Your Workflow

Run the workflow end-to-end and validate outputs in Sheets and Slack.

1. Click Execute Workflow to trigger Manual Run Trigger and process a sample row.
2. Confirm that Post Feedback Summary sends a Slack message with a score and flags.
3. Verify that Update Score Sheet writes Score, Flags, and LLM_JSON back to the correct row.
4. If the score is below 50, confirm Send Coaching Resources sends a training message.
5. Once verified, toggle the workflow to Active for production use.