The Canary in the Data Mine

Transaction monitoring for Square merchants. Connect your account. Get a weekly report showing exactly where you're losing money.

See How It Works Partner With Us

Built on Square APIs · Read-only access · Your data stays yours

Canary Weekly Report — Mar 24-30
$437 in potential losses found this week
Biggest driver: Your Tuesday and Thursday evening refund rate is 3.8x your store average. That pattern has been consistent for three weeks and accounts for about $280 of this week's total. The remaining $157 came from two post-sale voids on Wednesday.
Recommended action
Review the 6 flagged refunds with your Tuesday/Thursday closing staff.

What Canary flags

A voided sale 3 minutes after the customer left

A $300 payment processed at 2 AM — you close at 10

One employee's discount rate is 5x the store average

Refunds filed faster than a real return takes

Enterprise retail background

30 years. IBM 4690 POS systems, national chains, high-volume production environments. Transaction processing at scale.

Big 5 process capability

Enterprise-grade methodology and support practices, applied to the SMB market through automation.

Data integrity

If we flag something, we have the evidence. Every finding is backed by immutable, timestamped records.

How Canary works

Connect Square → Transactions flow in real-time → Detection engine evaluates every event → Patterns surface as plain-language insights → Evidence sealed and timestamped

Square webhook ingestion

Real-time event stream with HMAC verification

Detection engine

Configurable rules per merchant

Immutable evidence store

INSERT-only, hash-chained, trigger-enforced

Case management

Court-ready documentation

AI analysis

Pattern narrative generation

PythonFlaskPostgreSQLSquare SDKDocker

The tLog problem

IBM created the transaction log in 1986. Every POS system since uses the same model: a mutable record on a server someone controls. The log can be edited. It happens.

Canary's evidence store is different. INSERT-only. Hash-chained. Merkle roots anchored to Bitcoin. Records can't be altered after the fact.

The gLog

Canary is building the successor to the tLog — an immutable transaction record that works from one receipt to one billion.

Not blockchain for its own sake. A direct fix for a specific problem in how retail transactions have been recorded for 40 years.

Open source

The detection logic, evidence storage, and data handling will be public. Transparency is the standard. Execution is the differentiator.

Partner with us

Merchant partner

You run a Square operation. We need a merchant willing to connect their account, work with real data, and help refine the product. Any size. Read-only access. No commitment beyond trying it.

Get in touch

Team / service partner

We're building enterprise-level retail tooling for the SMB market — multi-location support, a canonical retail data model for back-office integrations, and detection at scale. Looking for people or firms who want to work in this space.

Get in touch
Received. We'll be in touch.