# Security: approval by action and run receipts · Bonggy

> Bots use only the permissions you connect, nothing customer-facing goes out without a person, every run leaves a receipt, and no training on your data.

Source: https://www.bonggy.com/security

A VP doing diligence on a vendor whose bots work with their team's data should walk away comfortable. Here's how we approach it.

### Approval by action

Nothing customer-facing (emails, posts, sequencer pushes, published content) goes out without a person approving it. Internal output (a brief in chat, a Slack summary) can run without approval if your team chooses.

### Bots only use the permissions you connect

You choose which tools each bot can reach. Bots are built to be read-only by default and write back only where a flow your team sets up allows it. Nothing beyond what you connect.

### Every run leaves a receipt

What the bot read, what it did, who approved it, and what it didn't send.

### Built to protect your data

Built for read-scoped permissions, encryption in transit and at rest, and no training on your data. We don't use it to train AI models, and you can ask us to delete your data at any time.

## Data protection

Built for read-scoped permissions, encryption in transit and at rest, and no training on your data.

## Compliance

SOC 2 Type II: on the path, not attained.

For documentation ahead of a pilot, email founders@bonggy.com.