We go to A LOT of trade shows. Badge scanners are very expensive. Is there a way to go from a picture of a badge at a trade show (usually just name, title, company name) to a row in a clay table? a lot of times we're getting over 100, so trying to figure out how to get the info off the picture and into a clay table.
Definitely want to hear back on how this goes as well!
Thanks for the responses. Going to work on building a prototype on this today 🙂 will report back!
Here's a clean, practical workflow to solve this effectively: Workflow Solution: You're looking at three core steps:
Badge OCR extraction (Name, Title, Company)
Linking Badge Data to Rep Name & Notes
Pushing structured data into Clay
Step 1: OCR & Parsing Your main issue is reliably extracting badge data. GPT-5 can do this, but it’s manual. Here's a better approach:
Use an OCR API:
Recommended: Google Cloud Vision or AWS Textract. Both reliably parse printed text from images, outputting structured JSON.
Reasoning: Automated, fast, scales well.
Backup Option:
Manual GPT-5 upload for small volumes (10–30 badges). Fine in a pinch, but won't scale long-term.
Example OCR output would be:
.png){kind=logo}
{
"first_name": "Adam",
"last_name": "Rutkowski",
"title": "Sales Director",
"company_name": "Example Corp"
}Step 2: Linking Data (Badge + Notes + Rep) You want badge details PLUS rep assignment and voice notes in one place. Here's how:
Recommended tool: Airtable or Google Forms.
Google Forms is quicker; Airtable is richer.
Ideal Airtable setup:
Field 1: Badge image upload.
Field 2: Rep name (dropdown).
Field 3: Voice notes (recorded via mobile voice-to-text directly into a text field).
When someone fills out the form, Airtable automatically stores these three items together. Step 3: Automating Integration into Clay You now have structured badge info (via OCR) and corresponding notes and rep assignment (Airtable).
Zapier / Make Integration:
OCR API extracts badge data → writes structured data back to Airtable.
Airtable row updates trigger Zapier to auto-sync data into Clay.
Clay Table schema example:
| First Name | Last Name | Title | Company | Rep Name | Notes |
|------------|-----------|-----------------|--------------|----------|-------------------------------|
| Adam | Rutkowski | Sales Director | Example Corp | Nick | Interested, follow-up Monday |
This automatically keeps everything cleanly linked together. Recommended Tech Stack (Long-term Solution):
OCR Engine: Google Vision API
Structured Form Capture: Airtable (badge upload, rep, voice-to-text notes)
Automation Platform: Zapier (connects Airtable + OCR API → Clay integration)
Why not just GPT-5? (Clarified)
GPT-5 is excellent but still manual (upload limits, human-in-loop).
For rapid trade-show workflows, automation and OCR APIs significantly outperform manual GPT-5 image handling.
Quick (Short-Term) Solution (MVP): If you want speed right now with minimal setup:
Upload images to GPT-5 (10 at a time), output structured text.
Paste GPT output into Airtable or Google Sheet.
Manually match notes/rep data from forms and push to Clay via CSV import.
My Recommendation:
For this customer scenario (trade shows, recurring use), go long-term OCR + Airtable + Zapier solution.
If immediate proof-of-concept needed today, GPT-5 works temporarily, but OCR-based automation wins hands-down long-term.
Let me know if you want detailed Zapier workflow templates or Airtable setup help.
i'm thinking just creating a form with
badge scan picture
rep name (drop down)
notes on the lead (voice to text memos)
the gap is pulling the info from the badge scan (name, title, company name) and maintaining the rep name and notes from the individual record all into one row to push to clay for further enrichment
can you tell me more?
interesting. another gap i'm seeing is how we marry the notes to the exported data.
gpt5 should work but you'll be limited to 10 images at a time so 100 contacts might take like 30 min to do correctly
i'd just upload the images 10 at a time to GPT and ask it to make a list of the people in the images and then at the end ask it to aggrigate into a single list so you can copy and paste into a spreadsheet
tried that. this is the issue i'm getting. could try a different model i suppose The link provided leads to a Google Drive sign-in page which restricts access to the actual image content. Without access to the image, the required fields (first name, last name, title, and company name) cannot be extracted. Therefore, all fields are returned as empty strings.
Adam R. Have not tested this, but I'm thinking this might work:
A picture of the badge still has to be taken, with the information shown clearly
Upload them into Google Drive or any type of publicity hosted location and generate a link to the image
Export the list of image URLs into Clay
Within Clay, I believe with a GPT 4o model and beyond, you could get Clay to analyse the image, and return you the details you require.
Might be worth to test it with 2-3 images for a start!