Use case · Zapier
Redact customer data before ChatGPT in Zapier.
A Zap that sends a new ticket, lead or email to ChatGPT sends the email addresses, card numbers and keys in it too. Two Webhooks by Zapier steps in front of the AI step send the text through the Vrymdor™ AI Node first.
The problem
A Zap maps fields from one app into the next. The ChatGPT step reads every field you map into the prompt.
- A help desk or form trigger maps the whole message into the prompt field.
- Customers type card numbers and Social Security numbers into free text fields, and nothing in the Zap checks them.
- A key pasted into a ticket or a chat message goes into the prompt as plain text.
With the AI Node
You use Zapier’s own Webhooks by Zapier: the first step trades your agent for a short lived access token, the second sends the text to your node.
- Add Webhooks by Zapier, event POST, to your Node Host URL followed by /v1/token, payload type json, with agent_id and agent_secret as data. Test it: the result has an access_token.
- Add Webhooks by Zapier, event Custom Request, method POST, to your Node Host URL followed by /v1/node/run, with the header Authorization set to Bearer and the access token from the first step.
- As data, send the action sanitize with your text mapped into input. For free text, a Code by Zapier step that builds the JSON is safer.
- Test it. The answer has "decision": "allow" and the cleaned text in output. Map output into your ChatGPT step.
Zapier runs in the cloud, so your node needs an HTTPS address Zapier can reach. The hosted node (Pro and up) is the simplest.
Try it on a support form entry
The AI Node hides email addresses, phone numbers, card numbers, US Social Security numbers, private keys, and the API keys and tokens for OpenAI, Anthropic, Stripe, GitHub, AWS, Google and Slack, plus bearer tokens and JWTs. Each one becomes a label like [REDACTED:EMAIL], so the model still knows something was there.
The ticket number and the name stay as they are. Masked here: EMAIL 1, CREDIT_CARD 1, SSN 1.
This demo uses the Playground's Local engine: public, well known detectors set to the types the AI Node masks by default, in the AI Node's [REDACTED:TYPE] format. In your workflows, the Vrymdor™ AI Node does the masking.
What it does not cover
Plainly, so you know where you still need care.
- Names and street addresses are not masked by default. On Pro and Business you can add your own masking patterns for details with a fixed shape.
- It works by pattern matching on known formats, with the Luhn test for card numbers. A detail written in an unusual way, such as a card number spelled out in words, can get through.
- It hides private details before the model reads the text. It does not check what the model answers, and it does not make an AI model safe in general.
Start free, upgrade when you grow.
Community is free for up to 2 agents and 10,000 requests a month, with a hard cap so nothing is ever billed. Pro is $29 a month for up to 15 agents and 100,000 requests, and Business is $99 a month for up to 50 agents and 500,000 requests, each with 14 days free. Enterprise is sized to your contract.