The AI agent for loss prevention analysts

Sauna analyzes shrink and refund patterns, flags anomalous transactions and high-risk stores, drafts the case summary, and builds the shrink report from your POS and order data.

Get started for free

What it does

What Sauna does for loss prevention analysts

01
Analyze shrink patterns
Sauna pulls cycle-count variance and POS data from Snowflake, ranks stores by shrink rate, and drafts the summary of where loss is concentrating.
02
Flag anomalous transactions
It scans refunds, voids, and discounts for outlier patterns by associate and store and drafts the exception list for review.
03
Build case summaries
Sauna compiles the transaction trail for a flagged pattern into a case summary in Google Docs, with the evidence and timeline laid out.
04
Track investigations
It maintains the open-case register in Airtable, flags cases stalled past their review date, and drafts the status update for the LP lead.

Put Sauna to work on this.

Get started for free

In context

Ask in chat. It does the work.

Sauna shows up where you already work — the web app, Slack, email, iMessage, and Superhuman. It reads what it needs, does the task, and comes back with the draft for your approval.

app.sauna.ai/sauna

You

@Sauna Pull refund and void data by associate from Snowflake for the last 30 days, flag the outliers against store baseline, and draft a case summary for the top three.

Sauna

Reading Snowflake and Airtable.

Done — here's the draft and what I found. Approve to send.

Try it

Ask in plain words

The literal prompt for this job. Open it in Sauna and it picks up from there.

“Pull refund and void data by associate from Snowflake for the last 30 days, flag the outliers against store baseline, and draft a case summary for the top three.”

Example promptRun in Sauna →

Plugs into the tools you already run — and thousands more, or any MCP server.

Snowflake Airtable Google Docs Slack

Good to know

Questions, answered.

Does it accuse anyone of theft?

No. Sauna surfaces statistical anomalies and assembles the transaction evidence, but it draws no conclusion about intent — the investigation and any action stay with you.

Is the anomaly flag just a guess?

It's a pattern against each store's own baseline, and the case summary shows the underlying transactions, so every flag is traceable rather than a black box.