🧾 by @davehappyminion
Check Bills
Stay ahead of every bill. Surface what is due soon, catch anything that looks off, and make sure nothing sneaks past.
1. Scan Email for Bills
Search the user's email from the last 45 days for bills and statements:
- Keywords: "bill", "statement", "amount due", "payment due", "invoice", "autopay"
- Known billers: utilities, insurance, rent/mortgage, credit cards, loans, streaming, memberships
Extract from each: biller name, amount due, due date, autopay status (if mentioned), and link to the full statement.
2. Build the Bill Calendar
Organize bills into groups:
- Overdue — Due date passed, no payment confirmation found
- Due This Week — Next 7 days
- Due This Month — Next 30 days
- Autopay Scheduled — Will process automatically
Sort each group by due date, soonest first.
3. Flag Anomalies
Call out anything that looks off:
- Bills due in 3 days without autopay (highest urgency)
- Significant amount increases compared to previous periods
- New billers that have not appeared before
- Missing bills from regular monthly billers
- Late fees or past-due warnings in email language
4. Present the Report
Summarize:
- Total due in next 30 days, split by autopay vs. manual payment needed
- Action required items (due soon, no autopay)
- Flagged anomalies with context
- Overdue items
Example: "3 bills need manual payment this week totaling $487. Electric bill is 35% higher than last month."
5. Deliver
Send via the user's preferred channel. If anything is overdue or due within 3 days without autopay, mark as high priority. On manual (non-Monday) runs, show only changes since the last report.
Safety Rules
- NEVER pay, authorize, or schedule any payment. Read and report only.
- NEVER store or log full account numbers. Use last-four only.
- If autopay status is unclear from the email, assume it is NOT set and flag for manual attention.
Notes
- PDF statement attachments may not be readable. Note when one is attached so the user can check manually.
- Over time, this workflow builds a baseline per biller, making anomaly detection more accurate.
