Turn customer calls into your best content ideas — automatically

A good customer conversation can give you product, sales, and content insights at the same time. Most founders capture the first two. This workflow makes the third one systematic.

Stack

• Fathom: Records the call and sends the transcript to Zapier.
• Zapier: Moves the data and runs the workflow.
• ChatGPT (OpenAI via Zapier): Finds useful insights, compares them with your content archive, scores the article ideas, and creates an outline.
• Jotform: Stores your published content archive and the content briefs for review.

Before you start: This workflow uses paid Zapier features. The ChatGPT (OpenAI) app in Zapier also requires an OpenAI API account with prepaid billing. A ChatGPT subscription does not include API usage.

Step 1 — Create the two Jotform forms

Start by creating two forms. The first is your content archive. The second is where the system saves new content ideas.

• Open: Jotform
• Click: Create
• Select: Form
• Select: Start From Scratch
• Select: Classic Form
• Name the first form: Published Content Archive

Add these fields:

• Title
• URL
• Reader
• Problem
• Promise
• Summary

Create a second form and name it: Customer-Led Content Queue

Add these fields:

• Call title
• Call date
• Source recording
• Customer type
• Questions
• Objections
• Stories and examples
• Exact customer phrases
• Content pillar
• Best article angle
• Closest existing article
• Closest existing article URL
• Evidence score
• Audience relevance
• Specificity
• Commercial relevance
• Novelty
• Decision
• Claims to verify
• Outline
• Status

Each form submission also automatically appears in Jotform Tables for easy review.

Step 2 — Load your existing content

The workflow checks new ideas against your published content. To do that, it needs your existing articles.

Create a CSV with these columns: Title, URL, Reader, Problem, Promise, Summary

Don't use titles by themselves. Two articles can have different titles but solve the same problem.

• Open: Jotform Import Data
• Click: Import Submissions Now
• Choose: Published Content Archive
• Click: Continue
• Download Jotform's import template.
• Match your CSV to the template.
• Upload the file.
• Click: Import

Each imported row becomes a form submission and also appears in Jotform Tables. Imported rows also count toward your Jotform submission limit.

Keep the archive up to date. If your publishing platform has a Zapier trigger for new posts, use it to add new articles automatically.

Only add Reader, Problem, Promise, and Summary if your publishing platform already provides them. Otherwise, add an AI step to derive them from the article or add them manually.

Step 3 — Start the Zap when a transcript is ready

This Zap starts after Fathom creates a transcript.

• Open: Zapier
• Click: Create
• Select: Zaps
• Choose app: Fathom
• Select event: New Transcript
• Click: Continue
• Connect your Fathom account.
• Click: Test trigger

Only send the meetings you want to process. An easy way to do this is to include a prefix to each meeting title. Here are some examples:

• CUSTOMER — Acme onboarding
• SALES — Demo with Jordan
• COACHING — Session with Priya
• INTERVIEW — Newsletter reader

Add a Filter step.

• Click: +
• Choose: Filter
• Select the meeting title field from the Fathom trigger.
• Choose condition: Starts with
• Add one condition for each prefix.
• Join the conditions with: OR

The Zap stops if the meeting title doesn't start with one of those prefixes.

Step 4 — Pull the content archive into the workflow

The AI compares new ideas with your existing content. To do that, it needs your content archive.

Add another action.

• Click: +
• Choose app: Jotform
• Select event: API Request (Beta)
• Select method: GET
• Enter:https://api.jotform.com/form/ARCHIVE_FORM_ID/submissions?limit=1000
• Replace ARCHIVE_FORM_ID with the ID of your Published Content Archive form.

Keep archive entries short. If there are too many to send in one request, use a search or vector retrieval system to send only the most relevant articles.

Step 5 — Turn the call into structured fields

Use ChatGPT to extract the information you want from the transcript.

• Click: +
• Choose app: ChatGPT (OpenAI)
• Select event: Extract Structured Data

Under Unstructured Text, include:

• The transcript
• The meeting title
• The recording link
• The Jotform archive response
• Your audience definition
• Your content pillars

Under Description, paste this (or something similar):

Analyze this customer conversation. Use the transcript as the only source for claims about the customer, their words, their experience, and their results.

Use the content pillars only to classify the conversation. Use the content archive only to check for overlap.

Extract the customer's:
*   Questions    
*   Objections    
*   Failed attempts    
*   Stories    
*   Examples    
*   Numbers    
*   Memorable phrases
    
Return an empty value for anything that is missing.

Assign one content pillar to the conversation. Compare the reader, problem, and promise with the content archive. Ignore titles when checking for overlap.

Generate three article angles. Score each angle from 1 to 5 for:
*   Evidence    
*   Audience relevance    
*   Specificity    
*   Commercial relevance    
*   Novelty
    
Return the strongest angle and its scores. Return exactly one Decision:
*   REJECT -- No useful, relevant, specific angle.    
*   UPDATE -- Matches an existing article.    
*   VERIFY -- Needs evidence the transcript does not provide.    
*   APPROVE -- No overlap, no verification needed, and both Evidence and Audience Relevance are at least 4.    
*   Otherwise, return REJECT.
    
Create a practical outline for APPROVE, UPDATE, and VERIFY.

Never invent quotes, numbers, examples, results, or customer details. Stick ONLY to what was provided.

Under Values to Extract, create outputs for the fields returned by the AI. Map Call title, Call date, and Source recording from Fathom instead of the AI.

Step 6 — Route the decision

Add: Paths by Zapier

Create one path for each possible decision.

Path 1: Approved idea
Set to: Continue when Decision = APPROVE

Path 2: Update existing article
Set to: Continue when Decision = UPDATE

Path 3: Needs verification
Set to: Continue when Decision = VERIFY

Inside every path:

• Click: +
• Choose app: Jotform
• Select event: Create Submission
• Select form: Customer-Led Content Queue

Map the Fathom fields and AI outputs to their matching fields:

• For APPROVE, set Status to Ready for review.
• For UPDATE, set Status to Update existing article. Also map the matching article title and URL.
• For VERIFY, set Status to Research needed. Map unsupported claims to Claims to verify.

Don't create a path for REJECT. The rejected ideas stop here and are not added to Jotform.

Click: Test step

Open Jotform Tables and check that every value is in the correct field.

Step 7 — Create a weekly review

This report reviews the briefs in your queue. Rejected ideas are not included.

Create a third Jotform form called: Weekly Content Review

Add these fields:

• Week ending
• Repeated questions and objections
• Source calls
• Top unused opportunities

Create a second Zap.

• Choose app: Schedule by Zapier
• Select event: Every Week
• Choose day: Friday
• Choose: Time of Day
• Add another action.
• Choose app: Formatter by Zapier
• Select: Date / Time
• Select event: Add/Subtract Time
• Use the current date.
• Subtract: 7 days

Next, add:

• Choose app: Jotform
• Select event: API Request (Beta)

Retrieve submissions from the Customer-Led Content Queue created after that date. Pull the Published Content Archive into the Zap as well.

Then add:

• Choose app: ChatGPT (OpenAI)
• Select event: Extract Structured Data

Ask it to:

• Find questions, objections, and failed attempts that appear in at least two briefs.
• List the source calls for each pattern.
• Keep one-off comments separate.
• Compare the opportunities with your published content.
• Return the three best opportunities that are not already covered.

Finally, add:

• Choose app: Jotform
• Select event: Create Submission
• Select form: Weekly Content Review
• Map the report fields into the form.

By the end of the week, you'll have a list of the best content opportunities from real customer conversations. You still make the editorial decision. The workflow just handles the first pass.